AI Solutions
An AI Sales Assistant That Qualifies B2B Leads and Drafts Follow-Ups
Grounded in CRM and product data, it helps a small sales team respond faster without losing the personal touch.
Overview
Kestrel Industrial Supplies sells fasteners, tooling and safety equipment to manufacturers and contractors. Its eight-person sales team received hundreds of inbound enquiries a week through web forms and email, and response times had slipped to more than a day for many leads.
The brief
The company wanted an AI assistant that would help, not replace, its sales staff:
- classify and score inbound enquiries against clear qualification rules
- draft a tailored first reply using real product and pricing data
- summarise an account's history before a call
- keep every AI-generated message under human review before sending
What we built
The assistant reads each new enquiry, extracts company, product interest, quantities and urgency, and scores it against rules agreed with the sales director. Qualified leads are created or updated in the CRM with a suggested owner.
For each lead it drafts a reply that references matching products, stock availability and standard lead times, pulled from the product catalogue rather than invented. Sales reps review, edit and send from their normal inbox. Before calls, reps can ask for a one-page account brief covering recent orders, open quotes and past issues.
Technical approach
We built this as an AI agent with a small set of tools: CRM lookup, catalogue search, stock check and quote history. Retrieval runs over a PostgreSQL vector index of product data. The model layer uses Anthropic Claude for drafting and summaries, with OpenAI models used for embeddings, behind an abstraction that allows either provider to be swapped.
Guardrails include strict grounding (no prices unless returned by a tool), a human approval step for every outbound message, and an evaluation set of 300 real enquiries used to test each prompt change. The CRM integration builds on the client's existing CRM software rather than replacing it.
Project facts
Client
Industry
Category
Duration
Completed
Services
Technologies
- Anthropic Claude
We build AI assistants, agents and document workflows on Anthropic's Claude models, known for long-context reasoning, careful instruction following and strong writing.
- OpenAI
We integrate OpenAI models into business software for assistants, document processing, search and content workflows, with grounding, evaluation and cost controls.
- AI & Machine Learning
Applied AI and machine learning for forecasting, classification, recommendations and document understanding, alongside large language models where they add value.
- PostgreSQL
PostgreSQL is our open-source database of choice for SaaS platforms and cost-sensitive systems, with strong reliability, JSON support and vector search for AI features.
- React
We build responsive, accessible React front ends for dashboards, portals and SaaS products, backed by clean APIs and a component library you can maintain.
- Anthropic Claude
Gallery
Screenshots & walkthrough
Draft reply review panel
Lead qualification queue
Response time analytics
Features
Key features we delivered
Lead extraction and scoring
Pulls company, product interest, quantities and urgency from free-text enquiries and scores them against agreed rules.
Grounded reply drafts
Drafts first replies using live catalogue, stock and lead-time data, never invented prices.
Account briefs on demand
Summarises recent orders, open quotes and past issues before a sales call.
CRM sync
Creates or updates leads and activities in the existing CRM with a suggested owner.
Human-in-the-loop approval
Every AI-generated message is reviewed and edited by a sales rep before it is sent.
Evaluation and monitoring
A test set of real enquiries and usage dashboards track accuracy and adoption after each change.
Case study
Read the full story
- AI Support Assistant for a B2B SaaS Company
How a B2B SaaS company deployed a grounded AI support assistant that resolves 42% of conversations on its own and hands the rest to agents with full context.
- School Group Digital Transformation
How a five-school trust replaced six disconnected tools with one school management platform and parent app, cutting admin time on attendance and fees by around 40%.
What our clients say
We used to lose half a day's takings every time a store's broadband went down. Since the new tills went in, outages barely register, and head office can change a price at lunchtime and see it on every shelf-edge promotion by the afternoon.
Solutions
Built on our solutions
- Lead Management Software
Lead management software that captures enquiries from every channel, scores and routes them instantly and tracks follow-up, so no lead goes cold and marketing sees what converts.
- CRM Software
CRM software for B2B and service businesses: pipeline management, contact history, quotes, tasks and forecasting, configured around your sales process with optional AI assistance.
- AI Business Automation
AI business automation platform that reads documents, classifies requests and runs multi-step workflows across your systems, with human approval where it matters.
Services
Services behind this project
- AI Agent Development
AI agent development for multi-step business tasks: agents that use your APIs and tools to research, update records and resolve requests, with approvals, audit logs and clear limits.
- AI Application Development
AI application development for document processing, classification, summarisation, semantic search and forecasting — embedded into new or existing business software with measurable accuracy.
- CRM Development
Custom CRM development for sales, account management and service teams — pipelines, customer records, automation and reporting tailored to your sales process and integrated with your systems.
FAQ
Frequently asked questions
Will an AI sales assistant send emails without human review?
Not in this project. Every drafted message goes to a sales rep to review and edit first. Fully automatic sending is possible for low-risk messages, but we recommend starting with human approval and measuring quality before relaxing it.
How do you stop the AI from inventing prices or stock levels?
The assistant can only quote prices, stock and lead times returned by tool calls to the catalogue and ERP. Prompts forbid unsupported figures, and an evaluation set checks for hallucinated values after every change.
More work
Other projects
- Manufacturing ERP Rollout
A custom ERP for a mid-sized metal fabrication manufacturer, connecting sales orders, production planning, inventory, purchasing and finance in one system.
- Fleet Logistics Tracking App
A driver app and dispatch dashboard for a regional delivery company, with live vehicle tracking, digital proof of delivery and optimised route sequencing.
- Clinic Appointment Mobile App
A patient booking app and clinic scheduling back office for a group of private outpatient clinics, with online booking, reminders and secure pre-visit forms.
Insights
Related articles
- AI Web Application Development: A Practical Guide
A practical guide to AI web application development: use cases, architecture, retrieval-augmented generation, choosing LLMs, security, evaluation, costs and delivery steps.
- ERP vs CRM: What's the Difference and Which Do You Need?
ERP vs CRM explained: what each system does, how they differ, where they overlap, when you need one or both, and how to integrate them without duplicating data.
- How Much Does Custom Software Development Cost?
What drives custom software development cost: scope, integrations, team, compliance and support, with indicative ranges by project type and tips for controlling your budget.
Exploring AI for your sales team?
Talk to our AI team about a grounded, human-reviewed assistant for your pipeline.
Talk to an ExpertTalk to our AI team about a grounded, human-reviewed assistant for your pipeline.