SearchOps Studio · Agents and automation

Agents that do the repetitive work, with a person approving what matters.

Most of the value in agents isn't conversation. It is the admin around the work: finding the answer in a document, pulling numbers from three tools, drafting the reply, updating the record. We build agents that work on your own data, inside the tools your team already uses, with logs and approval steps you control.

Projects from
£4,000
Typical timeline
4 to 6 wks
Pricing
Fixed
#operations
JM

Operations manager

Which branches had fewer enquiries than last month, and why?

Agent

Three branches are down. Hull is down 31%, and its booking form was offline for two days. York and Bradford are within their normal range.

CRM: enquiriesUptime logAnalytics 4

Draft, waiting for approval

Email to the Hull manager: the booking form was down from Tuesday to Thursday. Can you follow up the enquiries we missed?

Approve and sendEdit
Illustration

Sound familiar?

Does this sound like your team?

If two or more of these are true, it is worth twenty minutes to see what a build would look like.

The same internal questions get asked in Slack every day.

Staff write similar replies, summaries or reports from scratch.

Someone checks a tool every morning just to see whether anything changed.

You tried a chatbot and it made things up.

What we build

The parts of the build, in plain terms

01

Answers from your documents

Policies, price lists, procedures and past work, with every answer linked to its source so people can check it.

02

Answers from live data

Ask the CRM, the booking system or your reports in plain English, limited to what that person is allowed to see.

03

Drafts for approval

Replies, summaries and reports drafted by the agent and approved by a person before anything is sent.

04

Actions in your tools

Create a record, update a status, schedule a task. Every action is logged, and anything risky waits for approval.

05

Where your team already works

Slack, Teams, email, or a panel inside your own platform. Nothing new to learn.

06

Your data in Claude or ChatGPT

An MCP server so your team's own assistant can read your systems, with the same permissions as everywhere else.

Example builds

What a project like this looks like

Examples of the kind of build we do, to make the scope concrete. They are illustrations, not client case studies.

A services firm with 200 staff

Internal knowledge agent

Before

New starters ask the same questions, and the answers are spread across 300 documents.

What we build

  • An agent in Slack answering from the handbook and policies
  • Every answer cites the document and section it came from
  • Unanswered questions logged, so the handbook gets fixed

A property management company

Enquiry triage

Before

Tenant emails land in one inbox and are sorted by hand.

What we build

  • Each email classified, prioritised and routed to the right team
  • A suggested reply drafted from the tenancy and repair history
  • Anything urgent flagged straight to the manager on call

An operations director at a franchise group

Morning briefing

Before

Every day starts with checking five dashboards.

What we build

  • The agent reads the overnight numbers from each system
  • One message at 8am: what changed and what needs a decision
  • Follow-up questions answered in the same thread

Typical projects

What it usually costs

Three common sizes, priced by the same model as the estimator. Load one into the estimator to see how each answer moves the price.

Document assistant

Answers from your documents, inside Slack or Teams.

£7,500 to £11,000

ex VAT

Timeline
3 to 4 weeks
Discovery
£1,500
Care plan
Essential, £295/mo

Data agent

Answers from documents and live data in two tools.

£15,000 to £23,000

ex VAT

Timeline
4 to 6 weeks
Discovery
£1,500
Care plan
Standard, £695/mo

Working agent

Drafts and actions with approvals across three systems, with personal data.

£29,000 to £43,500

ex VAT

Timeline
7 to 10 weeks
Discovery
£2,500
Care plan
Standard, £695/mo

What you get

  • The agent, running in the agreed channels
  • A test set of real questions with expected answers, rerun before every change
  • Logs of every answer and every action
  • Usage limits and a monthly cost cap you set
  • Documentation and a handover session

What discovery covers

  • The jobs the agent should do, and the ones it must never do
  • What data it can read, and on whose behalf
  • Where a person must approve before anything happens
  • A test set of 30 to 50 real examples
  • A monthly running cost estimate

How it runs

From first call to running software

Fixed price per milestone, and you can see every task at every stage.

  1. 01

    Discovery

    1 week

    We meet the people who will use it, map the tools and data, and agree a fixed price for each milestone.

  2. 02

    Build in milestones

    4 to 6 weeks, typically

    Each milestone ends in something you can use on a preview link. A shared board shows every task, and you get a short video update each week.

  3. 03

    Launch and handover

    Launch week

    We go live, train your team, hand over the documentation, and fix anything that comes up in the first 30 days.

  4. 04

    Care plan

    From £295/month

    Optional. Hosting, monitoring, updates when connected tools change, and improvement days on the higher plans.

Rough quote

Price your agents and automation project

Answer a few questions and the estimate updates as you go. Send it when it looks right, and we reply within one UK working day.

Step 2 of 3 · Agent or automation

What should the agent do?

Pick everything that applies.

Which tools should it connect to?

Pick everything that applies. Each one is an integration we build and keep working.

What kind of data does it handle?
Where are we starting from?
When do you need it?

Questions about agents and automation

All prices are in GBP and exclude VAT, which is added at the prevailing UK rate.