Tradestack: Product, team and how it was built
Tradestack (AI agents for the quoting and back office of trade contractors, United Kingdom): Five named contractor customers, November 2026. Evidence grade E: no revenue figure was checked. Team: 3 co-founders in 2024; 2 founders and an average of 2 people employed in the year to March 2026. Launched September 2024; still running.Added 7 November 2026
- Outcome
- Five named contractor customers, November 2026
- Evidence
- E no revenue figure was checked
- Launch
- September 2024
- Status
- Still running
- Team
- 3 co-founders in 2024; 2 founders and an average of 2 people employed in the year to March 2026
- Outside funding
- Antler investment, 2024; amount not published
- Main channel
- Not stated
What is not verified
No revenue figure was found. It was looked for in the company's pages, Antler's post, the LangChain and Arcade.dev customer stories, and the two sets of micro-entity accounts filed at Companies House, which do not include turnover. Antler wrote of a first paying customer in October 2024 and LangChain of first paying customers in September 2024, without an amount. The five customers are named by the company itself. The accounts state current assets of £221,229 at 31 March 2025 and £58,367 at 31 March 2026, without showing cash or revenue. A reader could look for a funding announcement, a founder interview with figures, or the next accounts, due by 31 December 2027. This story is grade E: the business is real and its product, price and customers are described from public pages, but no revenue figure was checked. Aiolum shows it as an idea worth studying, not as proof that it makes money.
Tradestack is a London company that sells AI operators, its name for software agents that do office work for trade contractors: quotes, supplier invoices, payment chasing, engineer scheduling and site paperwork. An AI agent is a program in which a large language model plans the steps of a task and calls other software to do them, here job-management, accounting and email systems such as Simpro, BigChange, Xero and Outlook. The pricing page says the Growth plan starts at £1,000 a month and is "fully managed", so Tradestack connects the systems and runs the agents for the customer.
Three co-founders started it in 2024: Sarel van Baalen (CEO), who had worked in construction, Vaibhav Pandey (CTO), an AI engineer, and Alexandra (COO), named only by her first name in the investor Antler's post of 8 October 2024. According to LangChain, the first version was built in six weeks: a WhatsApp assistant that priced painting and decorating jobs from text, voice notes and photos, which Antler says produced a branded PDF quote and had won a first paying customer. By November 2026 the site sold to electrical, plumbing, heating and other contractors and named five customer businesses.
No revenue figure was found in any source, so this is a grade E story. The company's filed accounts give the team size: an average of 2 people employed in the year to 31 March 2026 and 3 in the period before, and the about page now lists only the CEO and the CTO.

How much revenue does Tradestack make?
Each row shows the evidence grade and the source numbers from the list at the end of this page. The grades are explained in How Aiolum verifies.
| Fact | Value | Grade | How we know | Sources |
|---|---|---|---|---|
| Named customers, November 2026 | 5: DirectShine, MEC Commercial, Podscape, M Towlers Services, Graceful Decor | C | Named by the company on its own site: two case studies and three testimonials by business owners. No customer count is published. | [1, 4] |
| First version | a WhatsApp assistant for painting and decorating quotes, built in 6 weeks | C | LangChain's customer story of 25 September 2024 says Tradestack launched it to a community of 28,000+ users and secured its first paying customers. | [7] |
| First paying customer | by 8 October 2024 | C | Antler's post. It also mentions partnerships giving access to 29,000 UK trades professionals and an activation campaign on "the P&D forum". | [6] |
| Company | Trades Technology Limited, incorporated 19 June 2024, active | A | Companies House, company number 15788945, nature of business 62012 (business and domestic software development). The site footer gives the same name and number. | [9, 1] |
| Founders | Sarel van Baalen (CEO), Vaibhav Pandey (CTO); a third co-founder, Alexandra (COO), in October 2024 | C | Antler's post gives first names only. The about page in November 2026 lists the CEO and the CTO, and the 2026 accounts name the same two people as directors. | [6, 3, 10] |
| Team, year to 31 March 2026 | average of 2 people employed (3 in the period from June 2024 to March 2025) | C | Micro-entity accounts filed by the directors, unaudited. The careers page calls it "a small team". | [10, 5] |
| Price, November 2026 | Growth from £1,000 a month; Enterprise custom | C | The pricing page states this; see the pricing table. | [2] |
| Outside funding | Antler investment, 2024; amount not published | C | Antler's post explains why it invested but gives no amount. | [6] |
| Systems it connects to | Simpro, BigChange, Xero, ServiceM8, WhatsApp, Gmail, Outlook, Teams, SharePoint, Google Drive | C | Home page. Arcade.dev's customer story of 16 September 2025 says the Outlook and Xero connections went into production within a few weeks. | [1, 8] |
| Balance sheet | current assets £221,229 at 31 March 2025 and £58,367 at 31 March 2026 | C | The micro-entity accounts state these totals; they do not show turnover, profit or cash separately. | [10] |
| Time from launch to the November 2026 status | 26 months | calc | Calculated: September 2024 to November 2026. | [7, 1] |
How to read this table: grade A is the Companies House register itself; grade C is a statement by the company, its investor or a vendor that sells to it, including the unaudited accounts that the directors filed. A "calc" value was computed on this page. No row is a revenue figure.
| Plan | Per month, billed yearly (US$) | Per month, billed monthly (US$) | Limit |
|---|---|---|---|
| Growth | from GBP 1,000 | fully managed AI operators; 30-day money-back guarantee | |
| Enterprise | custom | custom integrations, higher volumes, dedicated account management; pilot period |
The page describes Tradestack as "a services-as-software business": the price covers onboarding, integrations, support and continuous improvement, with no separate line items. The Growth plan has a 30-day money-back guarantee; Enterprise has a pilot period. No yearly price, seat count or usage limit is published, and the prices are not converted to US dollars here.
When did Tradestack launch, and how did it grow?
From a WhatsApp quoting assistant for painters to an AI back office for trade contractors in about two years. Each entry carries its grade.
- 19 June 2024ATrades Technology Limited is incorporated in England and Wales. [9]
- September 2024CThe first version, a WhatsApp assistant that prices painting and decorating jobs, is built in six weeks on LangGraph Cloud and launched to a community of 28,000+ users, with first paying customers. [7]
- 8 October 2024CAntler publishes why it invested: three co-founders, a multi-agent system that produces branded PDF quotes, and a first paying customer. [6]
- 16 September 2025CArcade.dev describes Tradestack as an agentic back office for UK contractors, with Outlook and Xero connections in production. [8]
- 3 June 2026CAccounts for the year to 31 March 2026 are filed: two directors and an average of 2 people employed. [10]
- November 2026CThe site sells "AI operators for trades" from £1,000 a month, according to its pricing page, and names five customer businesses. [1, 2, 4]
Why did Tradestack work?
Four causes, in the order they mattered. Each one is a reusable lesson.
One trade, one document first
The first product did one job for one trade: a quote for a painting and decorating project. Antler wrote that it generated line items, calculated material quantities, checked material prices online and produced a branded PDF quote.
Work inside WhatsApp
LangChain wrote that Tradestack chose WhatsApp as the main interface because of its wide use among people who are not comfortable with new software. Users send text, voice notes and photos, and the agents ask for clarification when they need it.
A person checks the unusual cases
In 2024 the team stepped in through Slack or LangGraph Studio when a request was unusual, such as a material not sold in the UK. In 2025 the agents paused and asked for approval when, for example, an approved price did not match the original quote.
Connect to the systems contractors already pay for
Instead of asking contractors to move to new software, the agents work inside their job-management, accounting and email accounts. The CTO told Arcade.dev that the company prefers four or five deep integrations to a hundred shallow ones.
How it was built
These details come from two vendor customer stories (LangChain, September 2024, and Arcade.dev, September 2025) and the company's own pages. They are grade C. [7, 8, 1]
- A hierarchical multi-agent system built with LangGraph: a supervisor node expands the user's request into a plan and hands steps to sub-agents that share one state.
- Deployed on LangGraph Cloud; LangSmith records each run (tracing) and runs node-level and end-to-end tests, which the team used to compare models.
- In September 2024 OpenAI's gpt-4-0125-preview did better than gpt-4o for the planning node, according to LangChain.
- A custom middleware handles WhatsApp messages that arrive while the agent is still working (double texting) and keeps a message queue.
- Arcade.dev handles the sign-in (OAuth) and permissions with which the agents act in Outlook and Xero, for example drafting an invoice without approving a payment.
- The home page says customer data is hosted in the UK and is never used to train AI models.
Tools named in the sources
Each one opens its page in the Tools tab, with every story that uses it.
Who tried this before Tradestack, and why did they fail?
Three companies in our database of failed startups tried to fix the way homeowners and contractors find, price and run building work, and failed. The values below are from that database and are not independently verified.
Pro.com
The promiseHome renovations made simple by matching homeowners with reliable contractors.
Why it diedIt could not scale a service that needed local people in every city, and better-known rivals took the market.
Kaodim
The promiseConnect homeowners in Southeast Asia with vetted contractors through quotes.
Why it diedIt charged contractors for leads that often did not become jobs, so contractors left and the quality of the service fell.
Katerra
The promiseDo the whole building process in one company, from design to factory-made parts.
Why it diedIt built factories before it had enough customers, then bought other companies and went bankrupt.
The failed startups come from our database of 1,749 dead startups, retold here in our own words. Its entries are AI-assisted summaries of public sources, so these cards carry no evidence grade.
What changed
Pro.com and Kaodim stood between the homeowner and the contractor and had to win both sides, city by city. Katerra tried to replace the contractors with its own factories. All three needed large numbers of people or buildings before the business could work.
Tradestack sells to the contractor and leaves the contractor's customers, staff and software where they are. Its agents work inside WhatsApp, Xero, Outlook and the job-management system that the contractor already uses, so it does not need to win a market of homeowners first. The open question is the managed setup: the pricing page says Tradestack connects and runs the agents for each customer, which needs people for every new contractor.
Can one person copy Tradestack now?
The small start, yes. The managed back office needs a team.
A quote assistant for one trade, used through WhatsApp, is small enough for one engineer who knows the trade to build. Tradestack built its first version in six weeks with LangGraph, according to LangChain. The 2026 product, with connections to accounting, email and job-management systems and a managed setup for each customer, needs several people and outside funding.
Where it is still open: Other trades and other countries with their own price books and materials: roofing, flooring, landscaping, and quoting in languages other than English.
- 1Pick one trade and one document that takes hours to prepare, such as a painting quote.
- 2Put the agent where the tradespeople already are: WhatsApp, voice notes and photos.
- 3Have a person review the unusual cases at first, and let the agent ask for approval when numbers do not match.
- 4Connect to the accounting and job-management software that the trade already pays for before building anything new.
What could go wrong for Tradestack?
The risks are the mirror image of the causes above.
What is not verified
No revenue figure was found. It was looked for in the company's pages, Antler's post, the LangChain and Arcade.dev customer stories, and the two sets of micro-entity accounts filed at Companies House, which do not include turnover. Antler wrote of a first paying customer in October 2024 and LangChain of first paying customers in September 2024, without an amount. The five customers are named by the company itself. The accounts state current assets of £221,229 at 31 March 2025 and £58,367 at 31 March 2026, without showing cash or revenue. A reader could look for a funding announcement, a founder interview with figures, or the next accounts, due by 31 December 2027.
Trust signals on the home page
Under "Trusted by leading UK trade contractors" the home page shows 11 logos. Three of them (DirectShine, MEC, Podscape) match a case study or a testimonial on the site; the other eight have no case study or quote. The return figures, "~10x Return on spend" in the DirectShine case study and "Contractors typically see two to three times back what they spend" in the questions section, are the company's own statements, as were Antler's 2024 figures of quotes "20x faster" and revenue "up to 50%" higher.
Software with a managed service around it
The pricing page says: "Tradestack is a services-as-software business. We handle the setup, connect your systems, and manage your AI workforce on an ongoing basis." The work is done by the agents, but each customer gets setup and ongoing management from the company, and in 2024 the team handled unusual cases by hand. The home page says inbound calls are "answered" and shows a call being transcribed; it does not say whether an AI voice speaks to callers or whether callers are told that AI is involved.
Competition and model providers
On this site, Tutti (a WhatsApp assistant for tradespeople) and AngebotFIX (AI quotes and invoices for tradespeople) sell similar automation to similar customers, and Fleety dispatches jobs for plumbing and HVAC companies; field-service software such as Simpro and BigChange could add the same agents. LangChain wrote in September 2024 that the planning step ran on OpenAI's gpt-4-0125-preview, and the agents run on LangGraph Cloud and Arcade.dev, so a change in those providers' prices or terms reaches Tradestack directly. The about page says it is in startup programmes that give compute credits and priority access to models.
The team and the market changed
Antler named three co-founders in October 2024; the about page in November 2026 lists two, and the 2026 accounts list two directors and an average of 2 people employed (3 the period before). The careers page calls it "a small team" and advertised an internship on 9 June 2026. The first customers were painters and decorators; the site's "Built for" list now names electrical, plumbing, heating and gas, air conditioning, fire protection, security and property maintenance.
Sources
The grade next to each source is the best grade that the source can support. "Read in full" means that the text of the page was downloaded and the numbers were checked in it.
- C Tradestack home page: AI operators for tradesRead in full on 7 November 2026. Source for the product, the integrations, the logo wall, the testimonials, the questions section and the company name and number in the footer.
- C Tradestack pricing pageRead in full on 7 November 2026. Source for the plans, the price and the services-as-software wording.
- C Tradestack about pageRead in full on 7 November 2026. Source for the founders' full names and roles, the press list and the startup programmes.
- C Tradestack case studies: DirectShine and MEC CommercialRead the case-studies index and both case studies on 7 November 2026. Source for two named customers and the return figure.
- C Tradestack careers pageRead in full on 7 November 2026. Source for "a small team" and the internship of 9 June 2026.
- C Antler, 8 October 2024, by Jed Rose: Why we invested in Tradestack: empowers trades businesses to quote 20x faster and win more jobsRead in full (downloaded by the pipeline on 7 November 2026). The investor's account. Source for the three co-founders, the first product and the first paying customer.
- C LangChain, 25 September 2024: How Tradestack launched their MVP in 6 weeks using LangGraph CloudRead in full. A vendor's customer story. Source for the build time, the WhatsApp design, the models compared and the launch to 28,000+ users.
- C Arcade.dev, 16 September 2025, by Shawnee Foster: From WhatsApp Message to Xero Invoice: How Tradestack Actually Ships AI AgentsRead in full. A vendor's customer story with quotes from the CTO. Source for the Outlook and Xero connections and the approval step.
- A Companies House: Trades Technology Limited (15788945), overview and filing historyRead the overview and the filing history on 7 November 2026. Source for the incorporation date, the status and the accounts dates.
- C Companies House: micro-entity accounts of Trades Technology Limited for the year to 31 March 2026 (filed 3 June 2026) and the period to 31 March 2025 (filed 24 December 2025)Read both filings in full (iXBRL versions). Unaudited, filed by the directors. Source for the average number of employees, the directors and the balance-sheet totals.
- C Database of failed startups, local exportLocal export. AI-written summaries, not independently verified. Source for the failed predecessors.
Run this company? Send corrections or a verified figure to aiolum@livyden.com