Open-source Knowledge Compiler for AI agents. Built in Rust. MCP. AI Infrastructure. TG: https://xitter.cf/t.co/MtIyMbZffa
Joined August 2026
- Tweets476
- Following103
- Followers736
- Likes968
Pinned Tweet
EKOS Technical Progress & Tokenomics 2.0
One engineer. 100 days. 508 commits.
And weβre just getting started. π
Could your software engineers work this effectively?
I ran ZAUTH on my open-source project, EKOS, for the second time.
π Aug 23 β Oct 19
β’ 126 β 508 commits
β’ ZAUTH score: 73 β 74
β’ Zero external code matches reported
β’ No high-risk security patterns flagged
4x the commits. Real engineering progress.
Hereβs what sits behind those numbers:
π§ Original Engineering
β’ fact_ledger.rs: A custom log-structured storage engine built for time-travel queries.
β’ semantics.rs: Business semantics synthesis with Git-blame-based hypothesis tracking.
β’ cross_system.rs: Cross-database identity resolution using Jaro-Winkler and Jaccard scoring.
π‘ Built with Safeguards
β’ Consistent Result<T> error handling
β’ Token validation on the setup server
β’ Defensive coding practices designed to reduce common attack vectors
β
Tested Where It Matters
β’ Unit and integration tests across major modules
β’ Time-travel edge cases and cross-database identity collisions
β’ AI agents using MCP cannot resolve conflicts autonomously. That decision stays with humans.
π Architecture Before Implementation
β’ RFC-driven development
β’ A modular Rust workspace spanning ledger, identity, CLI, EKL, and distributed components
This isnβt about chasing commit counts. Itβs about building infrastructure engineers can actually use to bridge legacy systems and AI.
And this is where solana:CwubepDFJndzSKFmAMAm9u8Xx3PrizAwSq8hcGimpump becomes part of the bigger picture.
Weβre building the technology first. At the same time, weβre exploring how the ecosystem can better recognize and reward meaningful contributionsβfrom development and testing to community participation.
Thatβs the thinking behind the ongoing Tokenomics 2.0 discussions: aligning the long-term interests of builders, contributors, and the wider EKOS community.
The goal isnβt just to launch a token around an idea.
Itβs to build something realβand develop an ecosystem worthy of growing around it.
More engineering. More transparency. More opportunities for the community to contribute.
EKOS is open source. Weβre building in public. And weβre far from done. π₯
π ZAUTH: zauth.inc/reposcan/scan/cmvfβ¦
π» GitHub: github.com/alexeyban/EKOS/
solana:CwubepDFJndzSKFmAMAm9u8Xx3PrizAwSq8hcGimpump #Rust #OpenSource #AI #DataEngineering
@IBM @Dell @intel @AnthropicAI @UberEng @micro1_ai @pentagoniac @databricks
π The numbers tell the story. Now, it's time to SHOW the world.
EKOS has delivered measurable results in its legacy-to-Lakehouse test. But we're not stopping at numbers on a screen.
π¬ We're bringing the entire process to life.
From analyzing thousands of Java files to validating real data against Oracle, we're preparing a full demo that puts the results front and center.
Code tells the truth. Seeing it happen makes that truth impossible to ignore.
π₯ The results are in. The full picture is coming. Stay tuned.
@ekosproject
π€ Made with AI
π A MAJOR MILESTONE FOR EKOS π₯
From legacy ERP source code to a reconciled Lakehouse.
What happens when enterprise data pipelines are built on evidence from the source systemβnot assumptions?
We put EKOS to the test on iDempiere, an open-source Java ERP running on Oracle.
π₯ THE RESULTS:
π 4,542 Java files analyzed
π 1,060 pages of source-system documentation generated
π§ͺ 165 tables tested end-to-end, with every Bronze table reconciled against Oracle
π 3,664 of 3,667 Silver tests passed
π οΈ 6 real bugs uncovered during testing
βοΈ Zero LLM calls from analysis through Silver
But the real milestone goes beyond the numbers.
EKOS compiled source code, database migrations, Git history, documentation, and the live Oracle catalog into an evidence-backed knowledge ledger.
That evidence informed the pipeline:
β Trino β Iceberg Bronze
β dbt-trino Silver
β Dagster orchestration
Incremental loading is selected based on source-code behavior. Deletes are detected through bucketed key comparisons, with safeguards that hold mass deletions for human approval. Silver tests derive from database constraints and the Java model.
Every pipeline decision carries its evidence.
The full 849-table run remains constrained by available memory, and 195 conflicts still await human review. Gold Mart drafting works; the remaining work is ongoing.
This is a major step toward making legacy modernization more traceable, testable, and grounded in how enterprise systems actually behave.
EKOS: Enterprise knowledge, compiled into engineering evidence.
#EKOS #DataEngineering #Lakehouse #OpenSource4o
@Oracle @OracleCloud @googlecloud @thatjeffsmith @SQLServer @Dell @IBM
Ekos Project retweeted
The real opportunity goes far beyond a single test. π
Enterprises spend enormous resources dealing with legacy systems, hidden bugs, and fragmented knowledge. Solving these problems creates real business value.
@ekosproject is turning enterprise knowledge into verifiable, reusable infrastructure β with growing potential for real-world adoption.βοΈ
Weβre also open to collaborating with individuals and companies who recognize this potential and want to help move it forward.π€π·ββοΈ
Real technology. Real problems. Long-term potential.π
@sama @GavinSBaker @cz_binance
@Cognizant @connor_mc_d @CompleteSkeptic @elonmusk @ericzakariasson @finkd @paulg @rauchg @saranormous @JensenHuang
@hwchase17 @romainhuet @jeff_weinstein @OfficialLoganK
π A MAJOR MILESTONE FOR EKOS π₯
From legacy ERP source code to a reconciled Lakehouse.
What happens when enterprise data pipelines are built on evidence from the source systemβnot assumptions?
We put EKOS to the test on iDempiere, an open-source Java ERP running on Oracle.
π₯ THE RESULTS:
π 4,542 Java files analyzed
π 1,060 pages of source-system documentation generated
π§ͺ 165 tables tested end-to-end, with every Bronze table reconciled against Oracle
π 3,664 of 3,667 Silver tests passed
π οΈ 6 real bugs uncovered during testing
βοΈ Zero LLM calls from analysis through Silver
But the real milestone goes beyond the numbers.
EKOS compiled source code, database migrations, Git history, documentation, and the live Oracle catalog into an evidence-backed knowledge ledger.
That evidence informed the pipeline:
β Trino β Iceberg Bronze
β dbt-trino Silver
β Dagster orchestration
Incremental loading is selected based on source-code behavior. Deletes are detected through bucketed key comparisons, with safeguards that hold mass deletions for human approval. Silver tests derive from database constraints and the Java model.
Every pipeline decision carries its evidence.
The full 849-table run remains constrained by available memory, and 195 conflicts still await human review. Gold Mart drafting works; the remaining work is ongoing.
This is a major step toward making legacy modernization more traceable, testable, and grounded in how enterprise systems actually behave.
EKOS: Enterprise knowledge, compiled into engineering evidence.
#EKOS #DataEngineering #Lakehouse #OpenSource4o
@Oracle @OracleCloud @googlecloud @thatjeffsmith @SQLServer @Dell @IBM
π A MAJOR MILESTONE FOR EKOS π₯
From legacy ERP source code to a reconciled Lakehouse.
What happens when enterprise data pipelines are built on evidence from the source systemβnot assumptions?
We put EKOS to the test on iDempiere, an open-source Java ERP running on Oracle.
π₯ THE RESULTS:
π 4,542 Java files analyzed
π 1,060 pages of source-system documentation generated
π§ͺ 165 tables tested end-to-end, with every Bronze table reconciled against Oracle
π 3,664 of 3,667 Silver tests passed
π οΈ 6 real bugs uncovered during testing
βοΈ Zero LLM calls from analysis through Silver
But the real milestone goes beyond the numbers.
EKOS compiled source code, database migrations, Git history, documentation, and the live Oracle catalog into an evidence-backed knowledge ledger.
That evidence informed the pipeline:
β Trino β Iceberg Bronze
β dbt-trino Silver
β Dagster orchestration
Incremental loading is selected based on source-code behavior. Deletes are detected through bucketed key comparisons, with safeguards that hold mass deletions for human approval. Silver tests derive from database constraints and the Java model.
Every pipeline decision carries its evidence.
The full 849-table run remains constrained by available memory, and 195 conflicts still await human review. Gold Mart drafting works; the remaining work is ongoing.
This is a major step toward making legacy modernization more traceable, testable, and grounded in how enterprise systems actually behave.
EKOS: Enterprise knowledge, compiled into engineering evidence.
#EKOS #DataEngineering #Lakehouse #OpenSource4o
@Oracle @OracleCloud @googlecloud @thatjeffsmith @SQLServer @Dell @IBM
You canβt understand a legacy ERP by looking at its database schema alone. π
This week, I put EKOS to the test on iDempiere, an open-source Java ERP supporting Oracle and PostgreSQL.
The challenge? Reconstructing the engineering knowledge hidden across code, SQL, database migrations, and business logic. βοΈ
π Here are the results:
π Java Code Analysis
β’ 4,542 files parsed, none skipped
β’ 763 tables and 16,660 columns recovered with business names
β’ 2,882 table relationships identified in code
β’ 3,207 SQL statements extracted, with parsing status explicitly recorded
ποΈ Oracle Migration Analysis
β’ Failed parses reduced from 5,317 to just 9
β’ 62,251 migration statements analyzed
βοΈ PL/SQL Analysis
β’ All 61 stored procedures and functions recovered
β’ 991 of 992 statements parsed
β’ One remaining gap: a GOTO statement
𧬠Migration History
β’ 2,154 changesets and 26,060 changes analyzed
β’ Zero unrecognized changes
β’ Reported conflicts reduced from 2,372 to just 3, of which 2 are genuine conflicts
π‘ Why does this matter?
Legacy modernization requires more than schema extraction. It requires reconstructing business semantics, relationships, SQL logic, procedural code, and how the system evolved over time.
π§ EKOS turns this scattered engineering knowledge into a structured, evidence-backed representation.
π Every recovered fact links back to its source file and line.
π οΈ No LLM calls. Just rule-based analysis.
π This work is currently available on a separate branch.
π Next step: Connect a live Oracle database and compare the running system against the code. Every mismatch becomes a finding traceable to evidence.
#EKOS #DataEngineering #LegacyModernization #ERP #Oracle #OpenSource
@Oracle @OracleCloud @googlecloud @thatjeffsmith
@SQLServer @Dell @IBM
@UberEng @sqldaily @MySQL
@iDempiere @alibaba_cloud
This is just the beginning. Imagine what becomes possible when legacy systems can be understood, tested, and modernized through verifiable evidence. π οΈ
EKOS is building toward a future where enterprise knowledge becomes a foundation for smarter, more reliable AI and data infrastructure. The potential is enormous.
@ekosproject solana:CwubepDFJndzSKFmAMAm9u8Xx3PrizAwSq8hcGimpump
Real progress isnβt just about building something new. Itβs about testing it against real systems, finding hidden bugs, and verifying that the results match the source.
EKOS analyzed 4,542 Java files, tested 165 tables against Oracle, and uncovered 6 real bugs.
Thatβs how you turn technical claims into measurable evidence.
medium.com/@alexeyban/from-lβ¦
The first update is here. π₯
The Oracle-to-Lakehouse work is progressing βοΈπ οΈ
@ekosproject | Building the Next Layer of Enterprise Data Infrastructure ποΈ
Moving enterprise data from Oracle to a Lakehouse is one thing. Preserving data integrity, validating transformations, and managing changes safely is another.
Thatβs exactly what weβre building at EKOS β and the progress is strong. Weβre advancing toward a more reliable, traceable, and controlled approach to enterprise data pipelines.
This work creates opportunities for broader conversations around enterprise infrastructure, interoperability, and potential collaborations across the ecosystems of companies such as @Oracle , @Google , @IBM , @awscloud , @Dell , and @databricks .
The bigger opportunity goes beyond a successful demo: turning solid engineering into meaningful enterprise adoption.
The results so far have been strong: the core pipeline is running successfully, validation has made significant progress, and EKOS is turning a complex enterprise data challenge into a practical, evidence-driven workflow.
π₯β³π€
You canβt understand a legacy ERP by looking at its database schema alone. π
This week, I put EKOS to the test on iDempiere, an open-source Java ERP supporting Oracle and PostgreSQL.
The challenge? Reconstructing the engineering knowledge hidden across code, SQL, database migrations, and business logic. βοΈ
π Here are the results:
π Java Code Analysis
β’ 4,542 files parsed, none skipped
β’ 763 tables and 16,660 columns recovered with business names
β’ 2,882 table relationships identified in code
β’ 3,207 SQL statements extracted, with parsing status explicitly recorded
ποΈ Oracle Migration Analysis
β’ Failed parses reduced from 5,317 to just 9
β’ 62,251 migration statements analyzed
βοΈ PL/SQL Analysis
β’ All 61 stored procedures and functions recovered
β’ 991 of 992 statements parsed
β’ One remaining gap: a GOTO statement
𧬠Migration History
β’ 2,154 changesets and 26,060 changes analyzed
β’ Zero unrecognized changes
β’ Reported conflicts reduced from 2,372 to just 3, of which 2 are genuine conflicts
π‘ Why does this matter?
Legacy modernization requires more than schema extraction. It requires reconstructing business semantics, relationships, SQL logic, procedural code, and how the system evolved over time.
π§ EKOS turns this scattered engineering knowledge into a structured, evidence-backed representation.
π Every recovered fact links back to its source file and line.
π οΈ No LLM calls. Just rule-based analysis.
π This work is currently available on a separate branch.
π Next step: Connect a live Oracle database and compare the running system against the code. Every mismatch becomes a finding traceable to evidence.
#EKOS #DataEngineering #LegacyModernization #ERP #Oracle #OpenSource
@Oracle @OracleCloud @googlecloud @thatjeffsmith
@SQLServer @Dell @IBM
@UberEng @sqldaily @MySQL
@iDempiere @alibaba_cloud
Real engineering. Measurable progress. π₯
Oracle dialect support, column-level data validation, and successful validation across the initial loaded tables.
The Oracle-to-Lakehouse pipeline is taking shape, with more progress to come.
π github.com/alexeyban/EKOS/coβ¦
βοΈReal Tech. Real Impact.
From PostgreSQL to Apache Iceberg.πποΈ
EKOS Lakehouse demonstrates a complete end-to-end pipeline, generated and verified.
Watch the full demo.π₯οΈ
π₯ youtu.be/EGpYzn_-fv4
Ekos Project retweeted
π A successful data transfer is not the same as verified data integrity.
Moving enterprise data from Oracle to a Lakehouse raises critical questions: Is numerical precision preserved? Are incremental changes and deletions detected? Can every extraction decision be reviewed and verified?
βοΈ In our latest article, we explore these challenges and share practical insights from @ekosproject testing.
π§ͺ And this is just part of the story.
Technical test results involving EKOS and another project will be published soon. We tested real data-engineering scenarios, including data validation, incremental loading, deletion detection, and safeguards against unapproved changes.
π The results will speak for themselves.
Stay tuned.
@Oracle @googlecloud @Microsoft @iDempiere @awscloud @Dell @IBM @stripe
@OracleCloud @connor_mc_d
@sqldaily @SQLServer @thatjeffsmith @rowantrollope
@hwchase17 @LipBuTan1 @nicbstme @satyanadella
Ekos Project retweeted
π EKOS UPDATE π οΈβοΈ
We're making EKOS more accessible and easier to evaluate.
π₯οΈ The latest Web Console update introduces:
πΉ Database source management
πΉ Connection testing directly through the interface
πΉ Clearer visibility into how live database structures compare with knowledge compiled into the EKOS Ledger
π’ Why does this matter?
This is an important step toward helping companies, decision-makers, technical teams, and other interested users explore EKOS's real capabilities and evaluate the technology through practical testing.
π What's next?
Free EKOS testing for everyone is on our roadmap. We're working toward making it possible for anyone to experience the technology firsthand, explore its capabilities, and see how it works with real enterprise data.
π― Our goal: broader access, greater transparency, and real-world evaluation.
EKOS β Making enterprise knowledge traceable, verifiable, and useful for AI agents.
#EKOS #EnterpriseAI #AIInfrastructure #LegacyModernization
solana:CwubepDFJndzSKFmAMAm9u8Xx3PrizAwSq8hcGimpump @ekosproject
Ekos Project retweeted
π₯ BIG NEWS FOR EKOS! π₯
Today, we added and locked approximately 700,000 EKOS tokens in liquidity! π
And we're not stopping here.
We're consistently working on:
π Adding and locking liquidity
π₯© Strengthening the staking pool
π₯ Buying back EKOS tokens
π THIS WEEK'S NUMBERS:
π₯ 1,000,000 EKOS β Buyback
π₯© 500,000 EKOS β Staking Pool
π 700,000 EKOS β Locked Liquidity
That's 2.2 MILLION EKOS allocated across these three initiatives this week!
π We're planning more exciting weekends ahead. This is another step toward building a stronger EKOS ecosystem.
BUILD. STRENGTHEN. GROW. π₯
@ekosproject @OMID_0909
$EKOS
Ekos Project retweeted
π οΈβοΈEKOS TECHNICAL UPDATE: HUMAN-CONTROLLED EXECUTION
Enterprise knowledge isn't ready for execution just because it has been extracted.
EKOS is advancing its review and governance workflow with three connected improvements:
πΉ Evidence-backed sign-off
Human approvals are bound to a specific state of the underlying knowledge. When relevant data changes, outdated approvals can be detected.
πΉ Execution gates
Downstream workflow stages can be blocked until the required approvals are valid. Human decisions become enforceable controls, not just recorded statuses.
πΉ MCP visibility & Lakehouse integration
AI agents can read sign-off status through a read-only MCP tool, while designated EKOS Lakehouse operations can enforce approval requirements when sign-off enforcement is enabled.
β‘ Performance improvement:
The review summary dropped from 24.5 seconds to 1.8 seconds in the reported iDempiere test environment β approximately 13.6Γ faster.
This is a step toward enterprise AI workflows where knowledge is not only extracted, but reviewed, traceable, and governed before downstream execution.
EKOS β Making enterprise knowledge traceable, verifiable, and useful for AI agents.
#EKOS #EnterpriseAI #AIInfrastructure #DataGovernance
@googlecloud @Microsoft
@IBM @awscloud @Dell @intel
@stripe @UberEng
Ekos Project retweeted
π₯ THE RESULTS ARE COMING.
The iDempiere evaluation is progressing, and we're finalizing the results.
What we've uncovered could spark some serious conversations across the AI and Oracle ecosystems.
Legacy systems hold far more knowledge than most people realize. We're putting EKOS to the test to reveal what's hidden beneath the surface.
The findings are taking shape, and the final update will be published soon.
AI is evolving. Enterprise knowledge is next.
@ekosproject
π€
@Oracle @IBM @Dell @nvidia
@googlecloud @Microsoft
#EKOS #AI #Oracle #EnterpriseAI #LegacyModernization
You canβt understand a legacy ERP by looking at its database schema alone. π
This week, I put EKOS to the test on iDempiere, an open-source Java ERP supporting Oracle and PostgreSQL.
The challenge? Reconstructing the engineering knowledge hidden across code, SQL, database migrations, and business logic. βοΈ
π Here are the results:
π Java Code Analysis
β’ 4,542 files parsed, none skipped
β’ 763 tables and 16,660 columns recovered with business names
β’ 2,882 table relationships identified in code
β’ 3,207 SQL statements extracted, with parsing status explicitly recorded
ποΈ Oracle Migration Analysis
β’ Failed parses reduced from 5,317 to just 9
β’ 62,251 migration statements analyzed
βοΈ PL/SQL Analysis
β’ All 61 stored procedures and functions recovered
β’ 991 of 992 statements parsed
β’ One remaining gap: a GOTO statement
𧬠Migration History
β’ 2,154 changesets and 26,060 changes analyzed
β’ Zero unrecognized changes
β’ Reported conflicts reduced from 2,372 to just 3, of which 2 are genuine conflicts
π‘ Why does this matter?
Legacy modernization requires more than schema extraction. It requires reconstructing business semantics, relationships, SQL logic, procedural code, and how the system evolved over time.
π§ EKOS turns this scattered engineering knowledge into a structured, evidence-backed representation.
π Every recovered fact links back to its source file and line.
π οΈ No LLM calls. Just rule-based analysis.
π This work is currently available on a separate branch.
π Next step: Connect a live Oracle database and compare the running system against the code. Every mismatch becomes a finding traceable to evidence.
#EKOS #DataEngineering #LegacyModernization #ERP #Oracle #OpenSource
@Oracle @OracleCloud @googlecloud @thatjeffsmith
@SQLServer @Dell @IBM
@UberEng @sqldaily @MySQL
@iDempiere @alibaba_cloud
π EKOS UPDATE π οΈβοΈ
We're making EKOS more accessible and easier to evaluate.
π₯οΈ The latest Web Console update introduces:
πΉ Database source management
πΉ Connection testing directly through the interface
πΉ Clearer visibility into how live database structures compare with knowledge compiled into the EKOS Ledger
π’ Why does this matter?
This is an important step toward helping companies, decision-makers, technical teams, and other interested users explore EKOS's real capabilities and evaluate the technology through practical testing.
π What's next?
Free EKOS testing for everyone is on our roadmap. We're working toward making it possible for anyone to experience the technology firsthand, explore its capabilities, and see how it works with real enterprise data.
π― Our goal: broader access, greater transparency, and real-world evaluation.
EKOS β Making enterprise knowledge traceable, verifiable, and useful for AI agents.
#EKOS #EnterpriseAI #AIInfrastructure #LegacyModernization
solana:CwubepDFJndzSKFmAMAm9u8Xx3PrizAwSq8hcGimpump @ekosproject
Imagine when everyone can test @ekosproject for free and see its real power firsthand. Weβre just getting started. The future looks BIG. π₯
Very soon β¦.π·ββοΈπ οΈ
Explore the implementation behind this update:
github.com/alexeyban/EKOS/coβ¦
Ekos Project retweeted
From Information Visualization to an Interactive Engineering Knowledge Workspace.βοΈποΈ
A few days ago, the focus was on viewing engineering information through a dashboard.π
Now, the experience goes much further.
The EKOS Web Console πποΈ
is evolving from a simple visualization layer into a more interactive environment for exploring structured engineering knowledge, analyzing relationships, tracing source evidence, and inspecting knowledge compilation pipelines.
π§ Whatβs changed?
π Workspace Intelligence
Monitor knowledge objects, relationships, evidence, storage, and workspace activity.
πΈοΈ Interactive Knowledge Graph
Explore connections between engineering entities and investigate how components relate across the knowledge structure.
π Evidence & Traceability
Trace knowledge back to its supporting source material, including file paths and line references where available.
βοΈ Pipeline Observability
Inspect compilation stages, execution status, and logs to understand how engineering knowledge is processed.
β±οΈ Scheduling & Automation
Explore supported scheduling capabilities for recurring engineering workflows.
π§ͺ Evaluation & Semantic Review
Review available evaluation results and semantic findings to assess the quality and consistency of compiled knowledge.
π οΈ Configuration & Control
Access project configuration and supported operations through the web console and its available tools.
ββββββββββββββββββ
The shift is clear: from simply viewing engineering information to actively exploring its structure, relationships, provenance, and processing lifecycle.
EKOS aims to make structured, evidence-backed engineering knowledge more accessible to AI agents through MCP.
The Web Console brings greater visibility into how that knowledge is organized, processed, and connected.
This is more than a dashboard upgrade. Itβs another step toward making engineering knowledge machine-accessible, traceable, and usable. π
π Explore the Web Console:
alexeyban.github.io/EKOS/preβ¦
π Open-Source Repository:
github.com/alexeyban/EKOS
π Official Documentation:
ekos.dev/
#EKOS #AI #Engineering #KnowledgeGraph #MCP
@OpenAI @IBM @Dell @UberEng @intel @OMID_0909
π₯ BIG NEWS FOR EKOS! π₯
Today, we added and locked approximately 700,000 EKOS tokens in liquidity! π
And we're not stopping here.
We're consistently working on:
π Adding and locking liquidity
π₯© Strengthening the staking pool
π₯ Buying back EKOS tokens
π THIS WEEK'S NUMBERS:
π₯ 1,000,000 EKOS β Buyback
π₯© 500,000 EKOS β Staking Pool
π 700,000 EKOS β Locked Liquidity
That's 2.2 MILLION EKOS allocated across these three initiatives this week!
π We're planning more exciting weekends ahead. This is another step toward building a stronger EKOS ecosystem.
BUILD. STRENGTHEN. GROW. π₯
@ekosproject @OMID_0909
$EKOS
