Automate anything

You can now build your own flows in HASH, automating processes without leaving the application.

As your HASH web builds itself, flows can be set up to:

• →

  enrich entities with additional information;

• →

  quality-check or further-verify information added to your web;

• →

  process entities according to an algorithm (e.g. implement "lead scoring");

• →

  synchronize information with your own systems or third-party applications (via built-in integrations, or using webhooks).

Click the "Flows" link in your left-hand sidebar in HASH to get started.

Draft updates

In a previous update, we introduced the ability for new entities to be added to a queue rather than created directly in a user's web.

We've improved this functionality by allowing proposed updates to existing entities to be held in this queue for review, as well.

Action queue

Suggested updates and draft entities can now be found in your inbox under the "Actions" tab, which contains a queue of all items in HASH requiring action. This replaces the Drafts tab in HASH's sidebar.

See drafts in the Browser Extension

The HASH Browser Extension now shows a "draft" icon next to proposed entities held in a user's actions queue, distinguishing them from those added directly to a user's graph.

Developer Docs

The HASH Developer Docs are now publicly available, and may be of interest to users of HASH with a technical background.

Our existing guide to developing simulations has been folded into these, and can now be found over on the hash.dev site.

We've written up a series of technical articles detailing just how HASH actually works, and what's coming up next. We're also calling for open-source contributors to HASH. You can read the first post in the series, and find out more, over on the HASH developer blog.

Self-Hosting HASH

• →

  Instance administration panel: a visual interface for administrating your server is now available to users self-hosting HASH. To access this, you'll need "instance admin" privileges.

• →

  Per-user feature flags: HASH feature entitlements can now be managed at a user level, in addition to through the setting of instance-level flags, allowing things like pages to be selectively enabled or disabled on a per-user basis, for testing and staged rollouts.

• →

  Support for Node 21: HASH can now be run in more recent Node environments.

User guide and website

A new HASH User Guide has been published, providing a helpful handbook for using HASH.

Our new website has also launched, which includes:

• →

  A refreshed homepage which explains what we're all about, including a video walkthrough

• →

  Improvements to the glossary, making it easier for new users to understand the core concepts underpinning HASH

• →

  A new pricing page that brings transparency to our monetization of HASH

• →

  Refreshed sign up and sign in pages (better than Google's)

Similarity search

We've restored the search bar in HASH, providing the ability to search for entities by name/value. We've also upgraded the "embeddings" used by entities in HASH to the latest OpenAI model, and are now generating embeddings for types as well as entities. This means that similarity-ranked searches should now be more accurate, and types can also be discovered/compared.

Entity labels

Properties expected on an entity can now be designated "labels", allowing them to be used as a means of uniquely identifying or otherwise quickly distinguishing entities of that type. The primary label shown will always be context dependent.

File analysis

Files uploaded to HASH will now be automatically processed, with information from DOC(X), PPT(X), & PDF files extracted upon upload. Files are as a result made searchable by their contents, and entities can be automatically identified within uploaded files, and created in your web.

Grid view

In addition to the existing table and graph views, entities can now also be viewed in a grid. This is especially useful for browsing file entities, and thumbnail previews will be shown for file types where available.

Other improvements

• →

  Enhancement: HASH now handles failed uploads more gracefully, providing a handy "retry" button.

• →

  Bug fix: Improved scrolling in extra-wide tables, fixing an issue where hovering over a cell that was not fully in view would cause the scroll to jump to the cell.

• →

  Bug fix: Addressed an issue that could cause the entities table to crash the page under certain circumstances.

Export to CSV

Entities can now be exported from tables in CSV format.

Improvements to entity inference

The AI entity inference capability introduced in the HASH Browser Extension has been made more powerful. Before creating new entities, HASH will now search for existing entities in your web to update or enrich, helping you progressively enhance existing entities, and capture relationships/connections between them.

Inference jobs that have already begun can now also be canceled (helpful for those times you may've forgotten to turn 'passive inference' mode off in the browser plugin, and realize you've a bunch of unintended jobs ongoing in the background).

More "failure cases" are now handled, which could prevent the AI from correctly inferring entities from a given webpage, and more entities can now be scraped from extra-large pages, well beyond previous limits.

When specifying new types of entities to infer, link types that appear on those entities will also now be added (along with the link destination's expected entity types).

New data types

Joining last month's newly added data types, Date and Date/Time have been made available for use in HASH.

Bulk process actions

Multiple drafts in the actions queue can now be accepted or ignored at once, speeding up the process of reviewing AI-inferred entities added to a drafts queue, and enabling more speedy growth of webs (where human review is still required or desired). As ever, HASH remains capable of creating and updating new AI-inferred entities directly in users' webs, bypassing the drafts queue, when human review is not required.

Other improvements

• →

  Enhancement: complex values, such as arrays of objects, are now previewable directly within the entities table in HASH.

• →

  Enhancement: add "Create new" button on the entities page

• →

  Enhancement: you can now filter in the entities table, and sorting has been improved

• →

  Enhancement: browser preferences are now persisted through login/logout, and across uninstall/reinstalls

• →

  Bug fix: duplicate entities could be shown in some views