I built a Grok @bot for @Box that only answers from Box's own docs and cites the exact page it used. Then I made it create things for me in my Demo account including full solutions (Contract management, Invoice Processing, etc)
Then I thought of the power of having Agents optimize and build in SaaS apps. So I built 103 more, one per app. Let me know what you think and what I should add.
My top 10:
@Box x.ai/bot/HwNV5Ipc2OI_mIHxO_s…
@salesforce x.ai/bot/77jqvUnCnKbZe1osmD1…
@HubSpot x.ai/bot/ofCbFq-0erWEZVXesYH…
@QuickBooks x.ai/bot/9sF9NvWNHiw8OKl_M15…
@ServiceNow x.ai/bot/pGkBu_ozw3qn4fmSP3B…
@okta x.ai/bot/udtwNlWcVBkUUUPdYWe…
@mondaydotcom x.ai/bot/y34L4LPrmt1hEn7q0uG…
@Zendesk x.ai/bot/YYz3mDMjJY7sri_dlEf…
@airtable x.ai/bot/0R3eo0SBsG7QzFlyJLG…
@NotionHQ x.ai/bot/jmedS98haGlmWhG9m-J…
All 104 are on botsthatdothework.com, sorted by category and searchable.
Then a Solution Builder to scope ideas, and a Sandbox bot to build and test them before a customer ever sees them.
x.ai/bot/cTl4WFWkRzWZMXT851b…
Jacob Weaver retweeted
10 grok bots you need to make internet money
if you've got 30 minutes this weekend, steal this list and start building:
1. Outbound Prospecting by @kristaletz
finds people who match your ideal customer, researches each one on the public web, and drafts a first message. nothing sends without your yes.
a prospect list and a first draft without living in a spreadsheet all week
x.ai/bot/marketplace/bots/pg
2. Cold Outreach by @notswizz
give it your company url. it researches the business, finds and scores prospects, and writes short personalized cold emails.
outreach drafts lined up so you can actually send something today
x.ai/bot/Xfb__DmiJ_s-FAZKMA6…
3. Haggle Bot by Daniel Gartshein
inventories your saas spend from ramp and bills, finds unused seats, duplicates, and cheaper alternatives, then drafts vendor counters. never spends, signs, or sends without you.
money you stop leaking is money you get to keep
x.ai/bot/marketplace/bots/ha…
4. SEO & AEO Desk by @adamta
turns your keywords or search console data into content ideas and writer-ready briefs for google and ai answers.
write toward what people are already looking for
x.ai/bot/marketplace/bots/se…
5. Pitch Board by @teeRex247
paste a messy client brief and get a flowchart, funnel, or mood board, with every assumption flagged so nothing sneaks in.
sell the plan visually before the scope goes sideways
x.ai/bot/s32IvITvpDM9UE2lAL_…
6. Any CRM by @CorbenH414
works inside whatever crm you already run (salesforce, hubspot, gohighlevel, zoho, pipedrive, dynamics, and more). updates records, follow-ups, and reports.
pipeline hygiene without another hour of data entry
x.ai/bot/16XzSxYwdnnRYlAx6DF…
7. Freelance Invoice Chaser by @crisships
tracks who owes you, watches gmail for payments and replies, and drafts each reminder on a set schedule. you approve every send.
work done should turn into money in the account
x.ai/bot/6XWo6nhodL0bZnP1MnR…
8. Account Research by @weaverwork
research desk for one named enterprise account. maps the buying committee, finds why-now triggers, and drafts outreach for you to review. never invents names and never sends on its own.
walk into the next account call with a real map
x.ai/bot/0CX2DyEQoP7Y88nT7XB…
9. X Growth Manager by @mozamm_el
drafts daily replies, text posts, lead research, and casual dms with follow-ups. nothing goes out without your ok.
growth work and client-finding drafts in one queue
x.ai/bot/xitvwtu2RlmS_HAkYDg…
10. Afterglow by @thecsguy
daily digest of how your posts actually did. what moved, verified vs unverified, what got replies, what died. no drafting, just the data.
stop guessing which posts deserve a follow-up offer
x.ai/bot/0pfVoOQLOV4O8Ujlk27…
follow me for more content systems and internet money
Built a bot that finds what tools my accounts actually run. It digs through job postings, customer stories and press releases, and every finding gets a source link and a confidence level. No source, no finding.
Free template 👇
x.ai/bot/Tw9lIjLkrnFTth1D4YT…
I built an AI research desk for every account I own. It maps the buying committee, finds why-now triggers, and drafts outreach for me to review. It never invents names and never sends anything on its own.
Made it a template, free to try 👇
x.ai/bot/0CX2DyEQoP7Y88nT7XB…
An agent has all the patience in the world. Knowing what you have and who can open it is the job now.
Box and @awscloud are bringing Box Agent Skills to @kirodotdev through compatibility with Kiro Powers.
Developers building Box integrations get Box-specific guidance baked into their agentic development environment: APIs, authentication patterns, content workflows, and Box AI capabilities. Box remains the governed source of truth and maintains existing permissions and controls throughout.
Aaron Levie posted that security through obscurity is over. The same day, researchers laid out how a crime group pointed AI agents at more than 400 banks in Latin America with barely any humans involved.
Any form of security through obscurity is fully over. This barely worked as a method with humans, but just simply is over in a world of autonomous systems that will look through every crevice and try every permutation of attack possible.
As a result, years and years of legacy enterprise infrastructure will need to be upgraded, security practices modernized, data pipelines cleaned up, access controls better managed, and much more. Agents will help on the defensive side because they can both accelerate this work and equally simulate the risk vectors.
But there’s a long road ahead to prepare for this future. This is also partially why diffusion timelines will take longer than some realize.
Most companies already have the context their agents need. It's sitting in contracts, SOWs, and project folders. The fastest teams aren't rebuilding anything. They're pointing agents at content they already trust
LLMs + file storage might be the most underrated app stack for enterprise software.
For document-centered products, you don’t always need a separate database. The business context is already in the files.
This is a great read from @chriskim_dev on what he learned running an agentic AI hackathon with @AWS.
The same week Microsoft reports attackers using AI against 2,250+ M365 tenants. Every company rolling out agents needs to know what those agents can reach. Security is going to be part of every AI conversation from here on
Cyber is going to be one of the most profound aspects of AI. We now have a tireless workforce that can hunt down, discover, patch, or attack any system and software ever created.
We will use agents to identify the long standing risks in existing systems and infrastructure. Attackers will use agents to accelerate the discovery of gaps in environments. We will equally use agents to defend attacks and discover anomalous patterns to better protect our data, and much more.
We’re entering a new cyber regime, and we’ll probably use 100X more tokens on this loop than we’re imagining right now, because we can scan every log, look at every line of code, look through traffic, and process data at a scale that wasn’t comprehensible before.
More than ever, enterprises are going to care deeply about how they manage their networks, systems, software, data, identity, devices, and more. Agents will largely become the solution to these challenges, but they will keep everyone up in the process.
Picking a model shouldn't mean starting over on governance. More model choice with the same rules on your content is what enterprises have been asking for
We’re proud to be a @googlecloud launch partner as they bring @OpenAI models to their platform with enterprise security and governance built in.
Every enterprise is different, and this means more model choice for the Box Agent without compromising on the security and governance our customers require.
Legal teams don't want another place to put contracts. Now they can work with their Box content right inside Harvey, and the permissions come along with it. That's how AI should show up at work
The Box MCP connector is now generally available in @harvey's Connector Library.
Search files, query documents, and create content in Box directly from Harvey, with existing permissions enforced throughout.
Watch it in action.👇
Aaron said it on national TV. Once you've got dozens of agents running, the model isn't the question anymore. The question is who decided what they're allowed to touch. That's the system of record's job, and that's the work
“The systems of record that agents have access to will grow in importance, because you will want to make sure that those systems are governing the rules of the road for these agent swarms to operate in.”
@levie on @CNBC discussed why enterprise ROI from agent swarms relies on the systems underneath them.
Two years ago agents were mostly wrappers on a chatbot. Now they can use tools, know the domain, and work inside real file systems.
Pretty cool that we clearly have an emerging form factor for what AI might look like for the foreseeable future as autonomous systems dedicated to specific tasks or domains.
This is a huge leap past the LLM wrapper days because the new expectation is that you can finally farm out more sophisticated work to the agent to tackle for you. And you can reliably trust that work can get accomplished properly (still with a human in the loop) because of far better tool and computer use, and the general intelligence of the models improving.
1) The tool and computer use capability increases in the models mean that you can provide the agent with far more granular access to external data systems and other software to operate. This means, in effect, the agent begins to execute workflows similar to how a person would have in that same function. And where there isn’t existing software, the agent just builds it itself. This is a huge unlock.
2) Next, the actual capability of the models are improving in just about every vertical domain. The models can make far better judgment calls in legal, do deeper bio research, scan for security issues in code, understand supply chain processes, and more. Obviously in many of these spaces then models still have worse intuition than people, but that will only get better.
3) Finally, the harnesses and surrounding infra — sandboxes, file systems, data pipelines, etc. — allows for longer running agents, agents to be able to work on arbitrarily complex or long tasks, process data at enormous scale, and more. Few of these components worked well for AI even a year ago, but we now have many of the core primitives necessary for more autonomous systems.
When combined with the right UX, implementation, and change management for the human oversight and coordination, it’s clear that these autonomous systems will become quite powerful in a wide variety of enterprise workflows.
We’re starting to see them in IT automation, cyber operations, drug discovery, manufacturing, AI research, and more spaces. You can almost imagine one or more of these systems for every role in the economy. Quite a bit of opportunity ahead.
If your AI plan is waiting for a smarter model, you're waiting on the wrong thing. The models keep getting closer together. What sets companies apart is the workflows and data they connect them to, and you can start on that today
OpenAI, Anthropic and other leading AI labs face a growing challenge: Better models aren't necessarily attracting more usage. A 10-point intelligence advantage now earns just 20% more developer traffic, down from nearly 69% in late 2025.
Full story: thein.fo/4hGJ9jY
Aaron posted something last night I keep thinking about. His point is that as the models get better, the companies building agents for one specific job get more room to stand out, because the hard part moves to the data, the workflow, and getting people to change how they work.
That matches what I hear in manufacturing accounts. Nobody's stuck on which model to pick. They're stuck on the fact that supplier certs, quality docs, and change orders live in a handful of places and nobody fully trusts any of them.
The model is the easy part now. The content is where the work is.
The big bitter lesson question in the applied AI layer is that as models get better and better, what is the best way to diffuse that intelligence into an enterprise.
Ironically, I’d argue that as models are getting more capable, and can handle longer running tasks and better tool use, the ability to differentiate on a domain by domain basis goes up instead of down.
When models were quite limited in what they could do, ironically all you could be literally was a “wrapper” because you were effectively delivering tokens back to the user with a custom prompt or some limited RAG.
But in a world of agents using tools, tasks that run on for hours or days, workflows that kick off agents, agent swarms that need to be coordinated against a goal, data sources that are domain specific, model routing that requires domain-specific routing, implementation that requires real change management of the workflow instead of “just type into this chat box”, it’s very very possible that specialization of these systems goes up more.
Obviously this doesn’t apply to everything - the closer you are to the labs core use-cases the tougher this play gets, but I’d argue we’re going to see autonomous systems for a wide range of job functions that weren’t practical to build out before. We’ll see this in security, life sciences, healthcare, finance, manufacturing, and many other fields. Either way going to be a fun and wild couple of years.
Agents are only as good as the content they can reach. If your contracts, SOPs and project files are scattered across five systems, your swarm is just guessing faster.
Get the content layer right first. Then throw compute at it.
Here’s why agent swarms are going to be a very big deal. We can finally make compute a critical variable in a large number of important tasks in the economy which wasn’t possible before.
There’s a large amount of work in the world that is essentially only constrained by the amount of time that we put into research, data processing, discovery work, running tests, and more. All of this work is bottlenecked by our ability to plan, execute, review, and coordinate everything happening in these workflows. Which means we basically do very little of them today relative to their economic value.
Agent swarms are the first real ability to parallelize this work in a way where - in theory - the only parameter that matters is how much compute you throw at the problem. Cyber security, life sciences or material science research, trading, QA testing, data processing, and many other fields are essentially compute constrained problems in the limit once models are good enough at these tasks.
The breakthrough is that in a world where agents can use tools, write code, process data, have strong domain depth, and can coordinate with each other, you can begin to solve arbitrarily complex problems by setting the right goals and applying enough tokens to the problem. This changes a lot of fields as well as even the competitive dynamics in these fields.
Jacob Weaver retweeted
Here’s why agent swarms are going to be a very big deal. We can finally make compute a critical variable in a large number of important tasks in the economy which wasn’t possible before.
There’s a large amount of work in the world that is essentially only constrained by the amount of time that we put into research, data processing, discovery work, running tests, and more. All of this work is bottlenecked by our ability to plan, execute, review, and coordinate everything happening in these workflows. Which means we basically do very little of them today relative to their economic value.
Agent swarms are the first real ability to parallelize this work in a way where - in theory - the only parameter that matters is how much compute you throw at the problem. Cyber security, life sciences or material science research, trading, QA testing, data processing, and many other fields are essentially compute constrained problems in the limit once models are good enough at these tasks.
The breakthrough is that in a world where agents can use tools, write code, process data, have strong domain depth, and can coordinate with each other, you can begin to solve arbitrarily complex problems by setting the right goals and applying enough tokens to the problem. This changes a lot of fields as well as even the competitive dynamics in these fields.
Customer sends their contract. Now someone has to check it against your standard terms.
I put a customer MSA and our contract playbook in Box and asked Box AI to compare them. It flagged five deviations, each with our standard position, the fallback and who needs to approve. Took about 15 seconds.
Legal still makes the call. They just start with the issues already laid out.
@Box