tfMakers
AI-native builders, working inside your business.
You know you need AI. Getting it built is the hard part.
Most businesses can see where AI should fit, but there is no easy way in: the in-house team is already stretched, off-the-shelf tools stall after the demo, and hiring an AI team takes months. tfMakers gives you one, with small pods of innovators, engineers, designers and automation specialists who join your team and build it with you, starting with a working prototype in two weeks.
What tfMakers is
A team that builds with AI, embedded in yours.
Most companies can see where AI would help. The invoices that get re-keyed, the reports somebody builds by hand every Monday, the support queue that answers the same twenty questions. What they are usually missing is people with the time and the skills to build it properly.
A tfMakers pod fills that gap. It works inside your business, on your problems and with your systems, and it builds things you keep: working tools and automations that your people use every day, rather than slide decks about what AI could do.
- Support copilot: Drafts replies for your agents
- Adviser portal: Tools advisers use every day
- Quote builder: Quotes from your price list
- Invoice matching: Matches invoices to orders
- Lead scoring: Shows who to call first
- Contract review: Flags the clauses to check
- Document search: Ask your files a question
- Booking engine: Live availability, booked
- Churn alerts: Spots customers at risk
- Email triage: Sorts and routes the inbox
- Pricing engine: Prices that follow the market
- Identity checks: ID verified at sign-up
- Order tracking: Every order in one view
- Approval workflows: Sign-offs that chase themselves
- Partner portal: Self-service for partners
- Fraud flags: Catches the odd one out
- Meeting notes: Summaries and next steps
- Recommendations: Suggests what to buy next
- Data pipeline: Clean data every morning
- CRM sync: Keeps your CRM up to date
Who it is for
tfMakers suits teams like these.
Operations teams doing too much by hand
Copying data between systems, chasing approvals, building the same report every week. If a job is repetitive and rule-based, there is a good chance it can be automated.
Leaders with AI on the agenda
The board wants to see AI working in the business, and the in-house team is already fully booked. tfMakers gives you the capacity without a hiring round.
Teams who have tried the tools already
Off-the-shelf AI got you a promising demo but never made it into daily use. We build the parts that connect it to your systems and your data.
Businesses with a long list of ideas
Plenty of things worth automating and nobody free to test them. A Sprint is a quick, low-commitment way to find out which ones are worth building.
What we make
Some of the things a pod can build.
Every business is different, so treat these as starting points. Each one is the kind of system a Sprint can prototype in two weeks and a Swarm can build into the business once it has proven itself.
- Operations
Document processing
Reads invoices, claims and application forms, checks them against your rules and files them in your systems, so nobody re-keys them by hand.
- Customer service
AI service agent
Answers customer questions from your own systems and policies, and hands anything tricky to a person with the context attached.
- Sales
Proposals and quotes
Drafts proposals and quotes from your CRM, price lists and past work, ready for someone to check and send.
- Finance
Reporting that builds itself
Pulls the numbers from the systems you already run into a live dashboard and a written weekly summary.
- Risk and compliance
Compliance checks
Reviews documents and records against your rules and the regulations you work under, and flags what needs a person to look at it.
- Knowledge
Internal knowledge assistant
Lets your staff ask questions of your policies, manuals and past projects, and answers with the source it came from.
- Ecommerce
Catalogue and content
Writes and checks product descriptions, specifications and translations across thousands of items, in your voice.
- Planning
Forecasting
Predicts demand, stock or staffing needs from your live data, and flags early when a plan needs to change.
- Operations
Workflow automation
Routes approvals, requests and handovers between your tools and your people, and chases whatever gets stuck.
How it works
A hands-on process, run with your team.
We follow the double diamond, and we work through every stage of it with you. It is close, hands-on work: we consult with your team directly, make the big decisions together, and once something is live, use what we learn to start the next round.
Discover
We spend time with the people doing the work, go through your systems and data with them, and map every place AI could help.
Define
We pick the problem worth solving first, agree with you what success looks like and decide how we will measure it.
Develop
We build working prototypes in two-week cycles and put each one in front of the people who will use it.
Deliver
Whatever proves itself gets built properly: connected to your systems, secured, tested and handed into everyday use.
Iterate
Then we go round again, reviewing the live system with you to decide what to improve or build next.
Three ways to work with us
Sprint, Swarm and Hive.
Most clients start with a Sprint, move into a Swarm once something has proven itself, and settle into a Hive when there is a steady stream of work. You can start at any of the three.
01 · Validate
Sprint
Two-week objectives, working prototypes.
A pod takes on one clear objective at a time and puts a working prototype in front of you at the end of each two weeks. The point is to find out quickly whether an idea is worth spending more on, using something you can actually click through.
Ideal for
- Testing an AI idea before committing a real budget
- Getting something tangible in front of the board
- Narrowing a long list of ideas down to the ones that work
What you get
- Something real to react to every two weeks
- A clear end point and a small commitment
- Ideas that do not work get dropped early, while they are cheap
02 · Realise
Swarm
Turn a proven prototype into something the business runs on.
Once a prototype has earned it, we put more people on it and build it properly: connected to your systems, secured, tested and handed into the everyday running of the business. The same people who built the prototype carry it through, so nothing is lost in a handover.
Ideal for
- A prototype that has already shown its value
- Work that needs to plug into systems you already run
- Getting a tool into production on a fixed timeline
What you get
- Prototype to production with the same team
- Built into the systems and data you already have
- Made to be maintained, with tests and monitoring from the start
03 · Sustain
Hive
An embedded team, working at a steady pace.
For businesses with a lot to build. A tfMakers team embeds with yours for the long run, works through a shared roadmap with you, and keeps shipping at a pace you can plan around. It is the most cost-effective of the three when the work keeps coming.
Ideal for
- Larger organisations with a long list of things to build
- Building AI capability without hiring a team for it
- Ongoing automation across several departments
What you get
- The lowest cost per piece of work of the three models
- A team that gets to know your business properly
- A predictable pace and a predictable spend
Why tfMakers
Innovators and engineers who ship to production.
There are plenty of people who can get an AI demo working. Getting it to run safely inside a real business, every day, takes a different set of skills, and Touchfoundry has been building and running production software for years.
Engineers first
tfMakers comes out of a studio that builds and runs production platforms, including the one behind Takealot's Personal Shopper programme. The AI is one part of a system, and we build the rest of the system too.
Connected to what you already run
We plug into your existing systems with proper access control, logging and testing, so the new tool does not end up living in a corner that nobody looks after.
Careful with your data
Data handling, permissions and review are part of the build from the first day. That matters more with AI than with almost anything else you could build.
Still around after launch
Touchfoundry runs what it builds. If something needs changing in six months, the people who built it are still here to change it.
Who we are making for
Already building with
Get in touch
Tell us what you want to make.
Pick the model you are interested in, or choose not sure yet and we will help you work it out. Someone from the tfMakers team will reply.