Skip to content
Claude
ChatGPT
Gemini
Microsoft Copilot
Grok
Perplexity
Mistral

AI
implementation support

Housecat's Fractional AI team sets up Claude, ChatGPT, Gemini, Copilot, Grok, Perplexity, Mistral and more with your team, then trains your people to run the workflows themselves.

How we work

  1. Set up AI workspace

  2. Connect data and systems

  3. Identify your pain points or growth opportunities

  4. Design and implement the right solution

  5. Test and iterate

  6. Onboard and train team

What this looks like

One example for each tool. Yours will start from your own work.

Claude

Claude

Support replies your whole team can send

The work
A customer asks for a refund on a discounted annual plan. Your support lead knows which policy applies and when a manager has to approve. A new teammate has to look both up.
What we set up
We set up a Claude project with your current policy and a few replies your lead would send again, then write the instructions together: cite the policy passage, flag a missing purchase date, leave exceptions for the manager. We try it on past tickets, including one from before the policy changed.
What you keep
Reps draft with the policy in front of them, and your lead knows which document to replace when it changes.

Get help setting up Claude for your business

Claude Code

Claude Code

A library upgrade nobody has had time to do

The work
An engineering team has been putting off a library upgrade because nobody is sure which services touch it. The knowledge of how to run the tests, which folders to leave alone and what a good pull request looks like lives in the senior engineers' heads.
What we set up
We write a CLAUDE.md for the repository with how to build and test it, the conventions the team follows and the places Claude Code must not change, and agree which commands it may run without asking. We try it on an upgrade the team has already done and compare its pull request with the one a person wrote.
What you keep
Upgrades and other chores arrive as small pull requests an engineer can review, and a CLAUDE.md that helps new engineers as much as it helps Claude Code.

Get help setting up Claude Code for your business

Claude Cowork

Claude Cowork

A monthly vendor spend summary from a folder of files

The work
Each month an operations manager gathers invoices, a spreadsheet export and a few email confirmations to work out what the company spent with each vendor, and which charges look wrong. The checks they run are not written down anywhere.
What we set up
With the operations manager we choose the folders Claude Cowork may read and the ones it may change, then write the instructions together: total by vendor, link every figure to the file it came from, flag duplicates and anything it is unsure about. We run it on last month's files and compare the summary with the one that was sent.
What you keep
A first draft of the summary in minutes, with a source for every number, and a short set of instructions the whole team can use and adjust.

Get help setting up Claude Cowork for your business

ChatGPT

ChatGPT

First-draft proposals that sound like your best partner

The work
A partner at a services firm rewrites a similar proposal every time a lead comes in. The pricing logic, and the tone of the proposals that won, live in the partner's head and in a folder of old documents nobody opens.
What we set up
We pick three or four proposals that won, write down the pricing rules and the phrases the partner always uses, and put them in a shared ChatGPT project with instructions: ask for the missing scope details before drafting, and never invent a price. We test it on last quarter's leads and compare the drafts with what was actually sent.
What you keep
A first draft in minutes that the partner edits instead of writes, and a short document the firm updates when its pricing changes.

Get help setting up ChatGPT for your business

Gemini

Gemini

A weekly customer update from your Google files

The work
Every Monday an account manager pulls numbers from a Sheet, notes from Docs and open questions from Gmail to write a customer update. Each customer gets a slightly different version, and the format lives in one person's head.
What we set up
With the account team and your Google Workspace admin we pick the folders Gemini may read, write a template for the update and a short checklist: numbers from the Sheet only, every claim linked to its source, anything unclear flagged. We run it on last month's updates and compare.
What you keep
A first draft in minutes that cites where each number came from, and a template the whole team can adjust.

Get help setting up Gemini for your business

Microsoft Copilot

Microsoft Copilot

Renewal briefs from the files you already have

The work
Before a renewal call, the proposal is in Word, the newer price is in an Outlook thread and an open question is in meeting notes. A useful brief says which price the customer actually saw, and where to check.
What we set up
With the account team and your Microsoft 365 admin we choose the source files, confirm pilot users have the right access and write a reusable request. For engineering teams using GitHub Copilot, we write repository instructions, practice on a recent issue and show reviewers how to check the diff.
What you keep
A brief reps can verify, and a repeatable way to teach the next rep and decide when the brief is ready for a call.

Get help setting up Microsoft Copilot for your business

Grok

Grok

A daily digest of what people are saying about you

The work
Your communications lead scrolls X every morning for customer complaints, mentions of competitors and news about your industry, and still hears about the real problems from a customer a day late.
What we set up
We write down the accounts, topics and keywords that matter, set up Grok to summarize them each morning in the same format, and agree what counts as urgent enough to go to a human right away. We run it against last month's posts and check what it caught and what it missed.
What you keep
A short daily digest with links to the original posts, and a clear rule for what gets escalated.

Get help setting up Grok for your business

Perplexity

Perplexity

Account research with sources attached

The work
Before a first call, a sales rep spends forty minutes reading a prospect's website, press and filings. Two reps researching the same company come back with different things, and neither can say where a claim came from.
What we set up
We write a research template with the questions your best rep always asks, set Perplexity up to answer them with a source for every claim, and add a rule for flagging anything it can't find. We try it on a handful of accounts you already know well and compare.
What you keep
A one-page brief per account that the rep can check in minutes, and a template the whole team uses.

Get help setting up Perplexity for your business

Mistral

Mistral

Contract summaries that stay inside your own systems

The work
A legal operations lead reads every supplier contract for the renewal date, the notice period and any clause that renews automatically. The contracts can't be pasted into a public chat tool, so the summaries get written by hand, one at a time.
What we set up
With the legal lead and your IT team we decide where Mistral runs and which contracts it may read, then write the instructions together: pull the five dates and clauses that matter, quote the page each came from, and flag anything it isn't sure about for a lawyer. We run it on contracts you have already reviewed and compare the results.
What you keep
A one-page summary per contract that a person checks against the source, and a clear record of where the data went.

Get help setting up Mistral for your business

Codex

Codex

A Friday report that runs itself

The work
An operations analyst rebuilds the renewal report from a billing export every Friday. The hardest part to hand over is a small rule in their head: does an account with two active contracts get one row or two?
What we set up
We pair the analyst with an engineer, turn a sample export and last week's finished report into a test, and document how to run it in an AGENTS.md file. Codex writes the script. The analyst checks rows and totals, and the engineer reviews the change.
What you keep
A script anyone can run, an example that explains the awkward case, and a clear way to change the rule when the export changes.

Get help setting up Codex for your business

Why Housecat?

  • Experienced team, experts in helping business teams automate workflows across email, CRM and other tools.
  • Tool-agnostic: we help you choose between Claude, ChatGPT, Gemini, Copilot and others, and we'll say so if a tool isn't the best fit for the job
  • Our technology platform provides the building blocks for rapid deployment of AI in your workspace
  • We build on real work, not demos: every setup starts from your own past examples
  • We ensure your team is confident to use AI as an important part of their day to day workflow

FAQ

Which AI tools do you work with?

Claude (including Claude Code and Claude Cowork), ChatGPT, Codex, Gemini, Microsoft Copilot, Grok, Perplexity and Mistral are the ones we set up most. We also work with Meta AI and DeepSeek, and with the AI features built into your CRM and other software.

We currently don't use AI. Where should we start?

We help you identify where AI could be most useful, help you get set up and introduce it to your team through a focused set of practical workflows. Your team learns using real examples from your business, so they can put it to use from the outset.

We already use AI. Can you still help?

Yes. We help you build on what's working, address gaps and make successful approaches repeatable across your team. That could mean improving your instructions, organizing business context or developing workflows for more complex tasks.

What kinds of workflows can AI help us streamline?

AI can help with tasks that involve reading, writing and bringing information together. Examples include drafting proposals, summarizing research, reviewing documents, preparing for meetings and turning notes into follow-up actions.

We can also help connect AI to the systems and tools your team already uses, so it can work with relevant information and support workflows across your business. We identify the best opportunities, then build and test workflows around your team's needs, with human review where it matters.

How do I connect other tools like CRM, email or Slack to AI?

The right approach depends on the tools you use and what you want AI to do. We help identify suitable integrations, set up access and permissions, and connect them to practical workflows. That could mean using your CRM to prepare for a customer meeting, finding relevant information in Slack or drafting an email based on previous conversations. We test the connections with your team and establish clear controls over what AI can access and which actions need your approval. We can also build custom connections to tools that you can't connect to AI natively.

Is it safe for AI to have access to business-critical data?

It depends on your setup, the sensitivity of the information and the controls in place. We help you assess what AI needs access to, review the relevant data-handling settings and policies, and configure permissions around specific workflows. We aim to limit access to what's needed and establish approval steps for actions such as changing records or sending messages. For particularly sensitive information, we work with your IT or security team to determine what's appropriate before connecting it.

How is this different from a training session?

Training helps people understand how to use AI. Implementation puts that knowledge to work in your business. We help design and test workflows, prepare the context AI needs and support your team as they use it. Training is included, alongside practical work on your actual processes.

We have struggled with getting AI to work reliably. Can you help?

Yes. We review examples of where things go wrong and identify what needs to change. That might involve clearer instructions, better source material, breaking a task into smaller steps or adding checks. We test improvements against your real work and help your team understand where AI is useful and where human judgment is needed.

How long does an implementation take?

It depends on the scope, the complexity of the workflows and how many people are involved. We agree on a timeline with you before starting, with clear milestones and time for testing and feedback. We generally recommend starting with a single team or workflow, so you can test an initial use case fast, often in the first week or two.

Bring us one workflow

Tell us who does the work today, which tool you're considering and what they have to fix by hand.

Talk to us

Stay updated

Get the latest updates on new Housecat features