Build n BloomPut ChatGPT and Claude to work.
Give your team a shared way to use the tools they already reach for. We set up ChatGPT or Claude around real tasks, company knowledge and the people who check the work.
25 minutes with a founder · no fee · no confidential files
For specialist advisory firms and other small and boutique businesses. Explore our advisory work. Australia-based, working remotely mainly with Australia and the United States.
A useful reply starts before the prompt.
ChatGPT and Claude implementation brings the whole task together: the source, the draft, the person checking it and the next step. Explore a customer-reply example, then change the source condition.
Ready for a person to decide. The draft has its source beside it. The customer lead checks the wording, edits it and decides what to send.
Read both paths
Complete information: customer request and approved policy → prepared draft with source → customer lead reviews → ready to send if approved.
Conflicting policy: customer request and conflicting sources → draft withheld → policy owner resolves the source → draft is prepared again for human review.
A fictional teaching example, not a client result or a live ChatGPT or Claude session. The handover does not send a message automatically.
Illustrative workflow. For an advisory application, see the assessment workflow and its fictional demonstration.
From setup to a team that can run it.
The written scope identifies the work you are buying. These are possible parts of it, rather than a bundle added to every engagement.
- Setup and access
- Choose an appropriate workspace and available plan controls; agree users, administration, sharing, permitted sources and responsibilities. Review provider terms and data handling for the actual use. We do not assume every plan has the same controls.
- Task design and configuration
- Prepare reusable instructions, suitable reference material, expected output and review steps. Use your current tools where they do the job. Custom code or an external integration appears in the quote only when it is needed.
- Tests the team can judge
- Check representative ordinary work, missing information, conflicting sources and an attempted use of material outside the permitted boundary. Agree what passes, what must stop and who decides whether an output is ready.
- Practice and handover
- Train people on the actual agreed task, including correction and recovery. Leave operating instructions, a named owner, a way to report problems and a working alternative if the tool is unavailable.
Keep the simplest setup that does the job.
If columns don't fit, scroll sideways. Keyboard: focus the table and use the arrow keys.
| Route | What changes | What we check |
|---|---|---|
| Improve the workspace | Configuration, task instructions, examples and practical training. | Whether the team can produce and review the agreed work without another system. |
| Add permitted knowledge | Organised reference material or a supported connection to selected sources. | Access, freshness, conflicting versions, source support and missing-information behaviour. |
| Build a connected workflow | A separately scoped integration or internal tool around preparation, review and handover. | Required read/write permissions, failures, human approval, operation and total effort. |
Subscriptions, workspace credits and API access have different terms. We check current entitlements and any applicable credits before quoting. Read the ChatGPT and Claude comparison for a task-based selection method.
Practise the work together.
The people who know the job help shape the implementation. They practise preparing work, checking its sources, correcting it and handling something that should not proceed. Useful habits should fit the way the team already works.
We assess preparation, review, correction and ongoing operation together. A faster first draft only helps when the rest of the work still makes sense. The aim is a more capable team with more room for the work and relationships it values; that improvement needs to be tested.
Before live use, agree which information may be processed, which providers and connected services are involved, who can access it and how access ends. Vendor privacy settings are one part of that decision. Read our security and governance approach.
A clear scope before you commit.
- Start with the conversation
- A free, 25-minute Fit Call with a founder. Keep a short reviewed provisional Fit Note whether or not you proceed; we agree its delivery time during the call. No preparation or confidential files are needed.
- Choose the useful next step
- A directly scoped Build or Connected Build when the boundary is clear. An optional paid Blueprint when the design needs work first: one important revenue path, three complete connected designs and a ranked first installation. A simpler route or no build may be the right answer.
- Accept work you can operate
- The quote records scope, fees, dependencies, date and written acceptance tests. Installation is separately commissioned. Thirty calendar days of in-scope stabilisation follow written acceptance. Continuing support is optional; there is no automatic retainer.
The Blueprint fee-credit and refund rules and installation remedies are set out in the full service guarantee. It covers the agreed deliverable and tests, rather than revenue, savings or adoption. See what an implementation quote should cover.
Questions about ChatGPT + Claude Implementation.
Do we need both ChatGPT and Claude?
No. We compare the work, the tools you already have, the controls you need and the effort of running each option. One product may be sufficient. Using both is justified only when the additional capability earns the extra cost and operating effort.
We already pay for subscriptions. What would you add?
Start with what works. The missing piece may be clearer task instructions, approved reference material, a shared way to check outputs or practical training. If a task needs information to move between systems, we separately scope that integration. Existing licences do not need replacing simply because we are involved.
Is this a training course or an implementation service?
Implementation can include configuring the chosen setup, preparing an agreed task, testing it and helping the team operate it. Training is tied to that work. If a short training or configuration engagement would be enough, we can quote that scope without adding a custom system.
Can ChatGPT or Claude use our company documents?
Where the selected product, plan and permissions support it. We first identify the permitted sources, authoritative versions, allowed users and what should happen when information is missing. Files, connected retrieval and a custom workflow have different operating needs.
Can you connect it to our other systems?
We can investigate an agreed connection to the tools your team uses. The scope depends on the available interface, access, required actions and tests. A chat workspace and an API-based integration are different arrangements; we check the current billing, entitlements and permissions rather than assume a seat includes them.
What does implementation cost?
We quote the agreed work before you commit. Workspace configuration, a repeatable task and a connected workflow involve different effort. The quote distinguishes our professional fees, third-party subscriptions or usage, your team's participation, acceptance and any continuing support.
Who owns the work after handover?
Your named operating owner and team run the agreed system, with training, written instructions and a way to report problems. Thirty calendar days of in-scope stabilisation follow written acceptance. Additional changes or continuing support are a separate choice and scope.
Make the next decision with a useful guide.
- Which tool fits the work?Use the same task pack and compare the complete effort, controls and practical fit.
- How do we bring the team with us?A rollout checklist that connects the setup, the task, practice and operating ownership.
- How should AI use company knowledge?Compare files, connected retrieval and a custom workflow, with source and access tests.
Is the question bigger than a tool rollout?
If several parts of the business need to change together, AI Transformation starts with the wider work, chooses a useful first change and builds from evidence. It may begin with ChatGPT or Claude, a tool you already use or a process improvement.
Bring one piece of work your team would like to do better.
Bring one part of the work you would like to improve or make possible. We'll help you understand where AI could be useful and agree the next worthwhile decision.
Free 25-minute Fit Call · keep a reviewed provisional Fit Note · timing agreed during the call