Concept demo · Fictional case

AI Mediation Platform
Case Closed.

Middlebot starts with unpaid invoices: private intake for each side, a shared proposal, and a voluntary way forward.

01 / PROBLEM

The cost of chasing the money.

For a freelancer, an unpaid invoice is more than an overdue number. It is time spent chasing, uncertainty about cash flow, and a working relationship slowly turning into a dispute.

Clients may have genuine concerns about delivery. Freelancers may believe they have done everything agreed. A £2,400 project can get stuck between those positions, while the effort of escalation erodes what either side hopes to recover.

The starting point: unpaid invoices with a defined scope and two parties willing to participate. If validated, expand into broader freelancer disputes and disputed deposits—not all dispute types at once.

02 / SOLUTION

A neutral first step.

Middlebot imagines private per-side intake: each party submits their account separately. The parties receive a shared proposal, not each other’s private submissions, and each decides independently whether to approve, counter or decline.

It would not decide who is right or force an outcome. Consent, a clear opt-out, and access to a human mediator would be central. The prototype uses scripted fictional content—not a live AI service. Its role views are a single-browser illustration, not secure access. There are no real uploads or submissions sent to a service.

03 / KEY NUMBERS

Hypotheses, not headlines.

Both sides

Agree to submit the case.
The primary validation measure—not one-sided interest or sign-ups.

Before making any growth or market-size claims, test the numbers that determine whether the experience is useful:

  • Cases both sides agree to submit: count a case only when both parties voluntarily agree to take part in the free concierge pilot.
  • Voluntary settlement rate: do both parties accept an outcome without pressure?
  • Human minutes per case: how much expert time is needed for a responsible process?

No results are claimed. These are the measurements for an initial manual pilot.

04 / WHY NOW

Better tools. The same human friction.

Language models make summarising statements and exploring options worth testing. But better language generation is not proof of neutrality, confidentiality or good mediation outcomes.

The inspiration is Tim Draper’s post about AI mediation . The focus on freelance invoice disputes is our extrapolation, not his endorsement of Middlebot.

05 / BUSINESS MODEL

Start with one case, not a subscription.

The proposed model is a flat fee per case, split by the parties. This is a pricing hypothesis, not a validated price or current offer. Test willingness to pay on both sides and how shared payment affects perceived neutrality. No checkout is included.

Start with unpaid invoices; consider broader freelancer disputes and deposits only after the initial workflow is validated. Operational costs, expert involvement, the fee split and what happens if either party opts out need review before charging.

06 / COMPETITION

The space is already being explored.

Dyspute.ai and Mediate AI are existing examples to investigate. Their inclusion is not an endorsement or a claim about their outcomes.

Other alternatives include direct negotiation, freelance-platform dispute processes, and human mediators. Middlebot’s proposed angle is private per-side intake, a shared proposal and a straightforward human handoff—not an untested claim of superiority.

07 / GO-TO-MARKET

Learn manually. Automate carefully.

Offer free concierge mediation in freelancer communities. Start with unpaid invoices and manually facilitate the process: separate intake from each party, then a shared proposal. This is a proposed pilot, not an active service or recruitment form.

Measure cases BOTH sides agree to submit. One interested freelancer is not a participating case. Record why the other side agrees or declines, voluntary outcomes and human minutes per case. Test manually before automating.

08 / RISKS

Trust is the product.

  • Neutrality: wording, proposed amounts and who pays can introduce real or perceived bias.
  • Confidentiality: real cases need genuine access controls, clear retention policies and explicit sharing consent.
  • Opt-out: either party must be able to stop without pressure to settle.
  • Jurisdiction: local mediation and legal requirements need qualified human review.
  • Enforceability: no generated summary should be presented as binding without appropriate review.
  • Suitability: power imbalances, threats or complex disputes may require a different process and human support.

This single-browser role-play cannot demonstrate secure access, legal compliance or the suitability of a real settlement.

09 / SUMMARY

A way forward worth testing.

The idea is simple: make it easier for two people to understand a disputed invoice and explore a voluntary resolution. The hard part is earning their trust.

Start with unpaid invoices, offer free concierge mediation in freelancer communities, and measure cases both sides agree to submit. Test a flat fee split by the parties later. Middlebot is a concept—not a substitute for legal advice.

Try the fictional case

A Half Baked concept · No real AI or legal service

Concept domain: Middlebot.ai

Search Middlebot.ai on Name.com

A naming idea only. We do not claim ownership or availability; the demo stays here.