The tag view groups articles by a concrete operational problem or implementation type, without exposing hundreds of raw frontmatter labels.
An AI agent for competitor monitoring watches named sources (competitor sites, pricing pages, news, job boards, tenders) on a schedule, summarizes only what changed, and links every item to its source. For a founder it starts from €1,200 net as a personal AI agent. The agent reports with a source, you make the call.
An AI agent for complaints handling acknowledges the case instantly, classifies it, assembles the context, watches the response deadline, and drafts a reply to your policy. A human always approves the substantive reply. Automation from €3,500 net, and the first step is a free process scan.
An AI agent for contract review does the first pass: it extracts parties, amounts, dates, termination terms, and penalties, compares them against your standard, and flags deviations with the exact clause quoted and located. A human decides. A production build starts from €6,000 net, and the first step is a free process scan.
AI agent examples across a business: email triage, customer service, complaints, quotes, invoices, reports, follow-up, receivables, competitor research, tenders, IT help desk, and questions about your data. These are typical process patterns, not client cases. Each example states what the agent does, where its boundary is, and where we break the process down in detail.
An AI agent for a construction company takes over the paperwork in the office: it assembles construction documents, chases subcontractor correspondence, monitors tenders, and reminds about receivables. Automating one process starts from €3,500 net, an agent running several from €6,000 net. A cost estimator always approves the estimate and margin. You start with a free process scan.
An AI agent for your database turns a plain-language question into a query against your systems and returns a number straight from the data, not from the model's memory. At Syntalith we build these from €6,000 net. Its safety is not a promise, it is architecture: read-only access, named metrics, and the query shown under every answer. You start with a free process scan.
An AI agent for email treats the shared inbox (office@, sales@) as a process: it classifies, extracts data, drafts replies for approval, escalates exceptions, and leaves a trail. Automation from €3,500 net, an agent that runs the process from €6,000. You calculate the hours triage costs on your own numbers. You start with a free process scan.
An "AI agent for Excel" is usually the wrong purchase, because Excel plays three different roles in a company and each has a different answer. Summing columns is handled by Copilot and Power Query. Feeding a sheet with data from email and systems is automation from €3,500 net. A sheet the whole operation depends on is a signal the process has outgrown Excel.
An AI agent for an internal IT helpdesk triages tickets, answers from a knowledge base with a cited procedure, and runs routines (password reset, folder access) within permission boundaries, escalating the rest to an engineer with full context. Simple automation from €3,500 net, a full agent that runs the queue from €6,000. You calculate the queue's cost on your own numbers.
An agency sells senior hours, then burns them on operations: briefs, reports, research, time tracking. An AI agent takes that operations layer (from €3,500 net), while creative, strategy, and the client relationship stay with people. You start with a free process scan.
An AI agent for a property developer takes over the repeatable part of the cycle: unit enquiries, transaction documents, project status updates, handovers, and instalment tracking. Automating one stage from €3,500 net, an agent that runs the process from €6,000 net. You set the boundaries and calculate ROI on your own numbers, starting with a free process scan.
An AI agent for a transport and freight forwarding company takes over the work on email and documents: order-inbox triage, extracting order data from PDFs into the TMS, statusing customers, assembling documents for invoicing, and receivables monitoring. Automating a single process starts from €3,500 net, an agent that runs a whole process from €6,000 net. You start with a free process scan.
Deployment is the start of the bill, not the end. AI agent maintenance means monitoring, updates when model providers change their APIs, token cost under control, rule fixes when your process changes, incident handling, and backups. A named engineer owns it, and the price follows the work, not the number of agents.
An AI agent has real advantages (round-the-clock work on text and data, consistency, scale without headcount, a checkable trail) and real disadvantages (hallucinations, maintenance cost, prompt injection, vendor dependence). The balance turns positive only with a narrow scope, a counted process, and boundaries. You start with a free process scan.
An AI agent for tender monitoring reads notices from BZP and TED, matches them to your company profile, and assembles a brief with deadlines. For many firms a free CPV alert is enough first. An agent pays off where there are too many notices to read by hand. You start with a free process scan.
AI implementation in a B2B trading or wholesale company starts not with an agent but with email orders, stock queries, and price lists. Automating one process runs from €3,500 net, an agent that runs the process from €6,000 net. You calculate ROI on your own numbers. The first step, a process scan, costs €0.
Do AI agents actually work? Yes, in a narrow, well-measured scope with boundaries and a trail. No, as the magical worker from the ad. Skepticism is justified: most 'AI agents' on the market are chatbots in new packaging. This piece gives you a tool to tell a working deployment from a dead demo.
Hermes Agent is an open-source (MIT) personal AI agent by Nous Research, created in July 2025. Its signature trait is the self-improving skills loop: it creates skills from its own experience and remembers across sessions. It runs on a server from about 5 USD a month, on any model, and talks to you on the channels you already use. It is a tool, not a ready-made guarantee of safety.
A multi-agent system is several specialized agents passing work to each other under an orchestrator. It genuinely helps with parallelism, per-stage specialization, and permission isolation, but it is a complexity multiplier: errors propagate, token costs rise, debugging is harder. Most companies need one well-built agent first. You start with a free process scan.
RPA and an AI agent are not the same thing. RPA is a software robot that replays clicks in user interfaces, good where there is no API and the process never changes. An AI agent reads content and decides within boundaries. Between them sits plain API automation, often the cheapest and enough. You start with a free process scan.