Developer builds citation-only sales knowledge base that admits when it lacks answers

·4 views

A developer created SalesWiki, an open-source starter kit for sales and marketing teams that stores compiled knowledge as typed Markdown files compatible with Obsidian. The system keeps raw evidence separate from conclusions and only returns answers backed by named sources, explicitly flagging gaps rather than generating unsupported responses. Changes to shared knowledge require a proposal and review step, keeping earlier evidence and the rationale for any update visible in Git history. SalesWiki is not a CRM or hosted SaaS product, but a governed knowledge base designed to help teams trace decisions back to verifiable evidence. A synthetic demo using a fictional Atlas Robotics deal illustrates how the tool surfaces reasons behind a suggested next action without removing human judgment from the process.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Why smarter AI models can cost more despite lower token prices

Developers using advanced AI models from OpenAI and Anthropic have noticed rising costs even as per-token prices fall, and the explanation lies in hidden 'reasoning tokens.' Both companies bill internal reasoning work as output tokens, even though this chain-of-thought processing is never shown to the user. A single visible 200-token response can carry a billed output of 12,000 tokens or more, depending on how much internal reasoning the model performs. Newer models also tend to reason by default, meaning an unchanged prompt can trigger significantly more token usage than it did on an older model. Both vendors expose reasoning token counts via API fields, allowing developers to audit actual usage and identify where vague or ambiguous prompts are driving up costs.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

How to Configure Sentry Source Maps for Self-Hosted Next.js Standalone Builds

Developers running self-hosted Next.js standalone deployments on platforms like Coolify lack the automatic source map uploads that Vercel's Sentry integration provides, leaving production stack traces unreadable. Without proper configuration, minified traces such as 's at chunk-a1b2c3.js:1:48213' offer no actionable debugging information. The @sentry/nextjs SDK can handle both uploading source maps to Sentry and removing them from the final build artifact, preventing exposure of minified code to the public. Proper setup requires configuring Next.js instrumentation.ts to initialise Sentry separately for Node.js and Edge runtimes, and exporting Sentry.captureRequestError to ensure request-level errors are captured. The build pipeline must be wired to upload maps during the build phase, before the container artifact is shipped to the runtime environment.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Alpine Linux Post-Install Security and Configuration Checklist for Servers

A developer has published a practical post-install checklist for Alpine Linux, targeting server deployments and embedded projects. The guide covers essential security steps including changing default root and user passwords, configuring SSH public keys, and disabling root login. It also addresses system hardening through kernel parameters, firewall rules via nftables, and disabling IPv6 where needed. Additional recommendations include enabling Linux cgroups for container workloads with Docker or Podman, setting up DNS resolution, and configuring logging tools such as rsyslog in place of the default busybox syslogd. The checklist references the official Alpine Linux Wiki and related documentation from Arch and Gentoo wikis for deeper guidance.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

What Is Unsupervised Machine Learning and Why Does It Matter

Unsupervised machine learning is a branch of AI where algorithms analyze raw data without labeled outputs or human supervision, uncovering hidden patterns on their own. Unlike supervised learning, which relies on pre-tagged examples to train models, unsupervised methods are especially valuable when labeled data is scarce, expensive, or unavailable. One of its core tasks is clustering, which groups similar data points together based on mathematical similarity measures. A widely used clustering technique, K-Means, works by iteratively assigning data points to the nearest centroid and updating those centroids until the groupings stabilize. The approach has broad real-world applications, including data compression, feature extraction, and discovering patterns that human analysts might otherwise overlook.

0 comments Read more at DEV Community